So, this question has both a specific example, and a general question. I address both the specific case (the first two headings) and the general case (the rest of the answer). Unfortunately, the answer to the general case is there is no answer that applies generally. There is no one-size-fits-all, this-is-what-a-GM-should-always-do-for-every-situation answer to this question. Every response to unexpected effects has to be addressed on a case-by-case basis that is specific to the table, the characters, the game, and the plot. Ultimately, that's the only possible answer anyone can offer.
First, the spell explicitly bars you from doing that
This isn’t an ex post facto houserule designed to fiat away your ingenuity; it’s literally just enforcing the existing rule that prevents this abuse.
Second, the cleric cannot summon orcas
The dolphin creature has specific stats, including a size and weight. It is somewhat generic, using the same stats for many related species, but ultimately what relates them is similarities in size, strength, behavior, and so on. The orca, though genetically a dolphin, is anomalous in most of those regards, being much larger. It is not appropriate to use the dolphin stats for an orca, nor is appropriate to treat the ability to summon a dolphin as the ability to summon an orca in particular.
Pathfinder, once again, explicitly says this. The orca is a separate creature from the dolphin, and the spell gets you a summoned version of the latter.
Third, adjudicating unforeseen, breaking uses of magic is one of the DM’s primary roles
Magic in 3.x is often vaguely defined and phenomenally powerful. It is common for magic to have unforeseen consequences, often greatly in excess of the expected power level for a given level. The DM’s judgment of such situations is often one of the most important reasons to have a human DM, rather than a computer that enforces the rules.
The DM should make these judgments almost always on a case-by-case basis, and the goal should be consistency and fun. Is dropping a dolphin, or a whale, or whatever, on someone “bad for the game?” I cannot tell you; it’s your game. The Pathfinder rules suggest that Paizo considered it bad for the game. 3.5 had the same rule, so Wizards of the Coast apparently felt the same way. But that doesn’t mean it can’t be a good thing for your game. It’s a bit silly, a lot cruel, and potentially does way more damage than you should be capable of, but those things may not matter, or even may be good things for your game. That is up to your DM, who, if he’s doing things well, should be trying to make all of the players happy.
Everyone seems to have really gotten a kick out of the plan, be really amused by the idea. The DM quashing it may have seemed unfun. But I will point out that if this becomes a regular tactic, and you’re able to crush your opposition with it, that will possibly make the game even more unfun. It also may ruin the mood and tone, which may also make the game unfun. Or else the DM has to redo a lot of work, which is very unfun for him, plus it means that effort isn’t going towards new things for you, which again reduces your fun.
So a DM has to make a judgment call on whether the problems an unforeseen interaction causes outweigh the enjoyment people get from it. This cannot be answered in a generic way, as every situation has different pros and cons. A DM definitely shouldn’t ban all, or even most, unforeseen effects: that’s what makes the game fun and interesting, and is a major source of reward for the players. To lose that would ruin the game. But eventually a line does need to be drawn, where the clever tactic, amusing though it is, is just not a good fit for the game.
Specific questions
Just for the sake of actually addressing each of your individual questions, despite their myriad false dichotomies. Please note that this is nothing like an exhaustive list of potential remedies! There are probably infinitely-many potential responses to these sorts of issues, all of them with their pros and cons and their situations where they are appropriate and their situations where they are horribly inappropriate. Which recourse is chosen depends on the group, the game, the characters, the situation, the work the DM’s already done, the things that have happened already within the game, and dozens of other factors that are impossible to list exhaustively.
- The GM can ban the spell.
- But then the players lose the intended functionality of the spell as well. If it's a particularly relied-upon spell (e.g. healing or resurrection) then the players are at a disadvantage for being creative.
A DM can, and should, ban a spell that cannot be used in a way except those that are bad for the game. In 3.x, there are quite a lot of those in my opinion, but that’s for my game, not yours.
- And what about NPCs/enemies? Do they lose the spell? If not, the players are at a disadvanteage. If so, won't the NPCs be unbalanced due to losing a spell they relied on?
An NPC statted with a given spell can always be modified to use a different one. Banning a spell that has already appeared in the game is usually not a great idea unless you explicitly ret-con (which is, itself, an extreme choice to be avoided, but is nevertheless sometimes the right choice).
- The GM can (sometimes) make the act impractical. Using the above example, the GM can simply rule that the dolphins are small and weigh very little.
- But sometimes the GM doesn't have that kind of wiggle-room. If our cleric were, for example, to summon a horse then we'd expect it to be the size and weight of a regular ride-able horse. And other spells might be even more specific.
- If the GM does make a ruling, this exception to the normal rules could itself be exploited. E.g. if the GM decreed our summoned horse weighs very little ("and is now impractical to drop on enemies, ha!") but it takes up the same volume, wouldn't it be extremely buoyant? So we couldn't drop the horses on people, but we could (as an example) use them to run across water?
As it turns out, horses are rather good swimmers, and have often been used to cross water.
Anyway, these examples are poor; it’s hard to imagine situations where these would be a good solution. But you can easily come up with more reasonable examples. For instance, setting the specific species of dolphin that you can get with the spell, to bar orcas (which are, in reality, already barred but for the sake of argument), would work without much in the way of “side-effects.”
But even with good examples, DMs should be careful about making such changes, because yes, side-effects are possible. That doesn’t mean they should or shouldn’t make such changes, just that they should do it carefully.
- The GM can threaten alignment penalties, having allies forbid the act, etc.
- This isn't always an option. If the usage of the spell doesn't violate your alignment in any way other then being unusual then it would be unfair to penalize the player.
Of course it’s not always an option. There isn’t any “one-size-fits-all” answer to these sorts of questions; the solution has to be tailored to the problem. Sometimes, this is the best option. Other times, it’s a bad option, or even not an option.
- Similarly, if the spell defeats the evil dark lord, why would allies forbid it? A mildly nasty act (sacrificing a defenseless dolphin) is certainly preferable to risking the lives of the villagers.
The latter point is extremely specific to a given person’s point of view. There are a lot of people who do not believe, or even categorically deny, that the ends justify the means. Paladins, for an obvious example. Also, how mildly-nasty that sacrifice is depends a lot on point of view, too; as a celestial dolphin, in particular, has an intelligence score that puts it above that of animals (and dolphins, themselves, are on the relatively high end of animal intelligence in the first place), that makes it a lot more dubious to consider the sacrifice “mildly” nasty. And then there are druids and the like to consider, who might treat any animal as sacrosanct and untouchable.
- The GM can ban the use of the spell "in that particular way", or simply say "it doesn't work."
- This is a frustrating cop-out, and won't stop certain types of players from experimenting with what they can get away with.
I don’t see that this option, as opposed to all the others, is specifically a cop-out. Particularly in this case, where it’t literally already a part of the rules; that’s not a cop-out at all.
That said, yes, there is a risk here. The DM should be aware of that, and weigh that against the problems that he’s trying to fix by doing so. There is no categorical answer here; sometimes, yes, he will have to go with the frustrating cop-out for the good of the broader game.
- E.g. if our cleric can't drop dolphins from 100m up, can he drop them from 10m? 5m? 1m? Can he summon dolphins on top of a battlement and then push them off?
The rules state that a dolphin has to be summoned into an actual body of water, of a size large enough for the dolphin to fit in. Seems a pretty fair, consistent, and reasonable rule to me.
Level 12 Analysis
Your AC is very low. At level 12, AC 18 is essentially a death sentence. I'm going to assume you have a source of Mage Armour from a party spellcaster, giving you an AC of 22, and a further Ring of Protection or Amulet of Natural Armour or Barkskin or something to up that by +2 to 24. 24 AC is still incredibly low.
The lowest attack bonus I found with a quick peruse of CR 12 monsters is +18, the secondary Bite of an Athach. That hits you on a roll of 6. The most common number I found was +23, which would hit you on a 2. I saw lots of +26, +28 and even a +33. Those all hit you on everything but a natural 1.
In other words, your AC is too low if anything targets you for you to survive. The DM will have to pull punches not to kill you.
Valiant Stand improves this. By not by enough. Enemies are still going to autohit you. Best case is +4 stat boosters on dex and wis, mage armour, nat armour/deflection bonuses, and valiant stand - 18 base, +4 from enhancement to stats, +4 armour, +2-+3 nat armour/def, +3 valiant stand = 32. That's a hit on a 9 from the lowest common attack bonus I found on CR 12 monsters.
Your Attack Bonus Is Pretty Decent, For a Monk. You have +16 to hit without Smite. With Amulet of Mighty Fists +2, that's +18ish to hit. With Belt of giant Strength +4 or Bull's Strength, that's +20 to hit. Everything that has AC that is supposed to be a defense has 26+, which is good - you hit most of the time. Power Attack and Flurry are counterbalanced by Valiant Stand, which you should be endeavouring to have always active by use of Stealth, which you don't have maxed out, uh, why? Popping up out of nowhere for a flatfooted smiting power attacking flurry should be your opening round action in every combat. Even your iteratives have a decent chance to hit, with this hit bonus. Only decent, though.
Your Damage Isn't Great. 1d10+1d4+5 (where is that errant +1 damage coming from?) averages out to 13 points of damage per strike. With Power Attack, that's 19 damage. With Smite, against evil or chaotic outsiders/dragons, you get /+18/, so 34 damage per strike, which is actually level 12-worthy. Too bad you get 3 total attacks per day with that damage. Sweeping Smite carries the Cha-to-hit bonus, but not the damage from the 'first hit'. It's kind of like a really terrible whirlwind attack that way.
Most of the time you'll be attacking for 19, on average. Some stuff is immune to bleed, or has DR/cold iron and good, etc, so it might be less. If you move, you only have one attack, so your first round barring stealth will nearly always be one attack at best. Rounds after that, you can flurry for four attacks, each which might hit for 19 damage.
So a max damage on an ideal round of around 34+19+19+19 damage, or 91 damage in 4 hits. That's not terrible. It relies on getting a full round action, though. It also relies on hitting with iteratives, DR of the kind you can get past, no miss chances etc. Most monsters at this level have at least 150hp, so you won't be winning in hp races. But you will be able to contribute a bit at least to the usual team of four on one beatdowns.
Your HP gets eaten in a round. With your AC, most CR9+ melee monsters are going to kill you to dead in a round. 11 Constitution on a melee combatant isn't a great idea, but then, you're a monk and need wis and and dex and str and also cha for being a paladin, so eh.
Your Saves are decent. They could be better, but they could also be worse. Most DCs seem to hover around DC 23 at this level, going up as high as 28 but also as low as 21. With your +11/+11/+13, you should have a 50% shot at most saving throws, and although fear and poison aren't exactly the most common life-threatening things to save against, it's still nice.
Power Level: Fighter.
This character can be relied upon to be roughly as good as an unoptimized fighter. The DM will have to pull some punches, and tough encounters (CR +2 or greater) will have a good chance of one-shotting this character by accident. In a party with Rogue or better power levels, this character will feel notably weaker than the rest of the group.
Note
Evaluating something without gear is hard, because at the lower end of the scale, gear can have more effect on your ultimate spread of numbers than the classes chosen. If your gear is random stuff, you will be weaker. If you plug it all into boosters for your FOUR necessary stats (dex, wis, str, cha), rings of deflection and other stuff that helps boost your melee stats to make you 'great' at that rather than 'mediocre to good', which includes buffs from friendly spellcasters, you will be stronger.
Note
This build is stronger at lower levels, where a handy Mage Armour plus your defensive stats can put you out of reach of opponents attack bonuses, and things are easier to hit and easier to kill. Level 12 is definitely where things are starting to go 'downhill', and they will likely get worse as enemy AC and AB increase and every enemy has flying or spells and abilities.
Note
The build has no movement capability other than skills-based climbing, swimming, jumping. I'm assuming again allied spellcasters will be patching that hole with buffs, especially since with the other gear i've assumed in your build you don't have the money to buy gear to get flight or teleport or whatnot.
Suggestions
Your main problem is that you're playing a monk. Without serious optimization, monks are very weak. You've done a bit of optimization here, and pushed it up to unoptimized fighter level, but fighter is still not great if you're planning to be in a party with casters. Or if you're fighting caster-style enemies like many demons are.
Most of the tricks I know to make melee characters more competitive are in 3.5e. I'm assuming there's no backward compatibility in your group. So that's all out the window.
Paladin isn't giving you much other than flavour, either. You've got a decent Flurry and Attack Bonus, but you're way too fragile to make good use of it, unless your enemies come in the 'always attack the guy using the full defense action with the tower shield' variety.
In this case, i'd recommend Summoner. Synthesist Summoner. It slices, it dices, it buffs, it has level appropriate abilities, natural attacks, the works. It is, in fact, so good that you could mix it with a Paladin dip and still be relatively awesome.
Evangelist lets you keep advancing as a Summoner while also adding a divine feel to the class. It's also the only prestige I can find that advances your eidolon aka living armour, so.
Other. than. that... there's a good rogue-based fear build, but I don't know how relevant that is to your concept.
I feel like there's some way, between Fighter archetypes and the Hungry Ghost Monk monk archetype, to create kama-wielder who reaps ki from his opponents and uses it to chain-stun them with a huge Wisdom and Dex, just whittles them away with flurries of weak, ki-reaping attacks and then spends the ki on stuns. There's too many damned archetypes though.
Even if you're taking summoner, it's kind of worth dipping monk as well. You have to be unarmoured to use your synthesism, so you could benefit from a high wisdom score to your AC. Not worth going Hungry Ghost, because the ki-drain only comes at level 5. Just a one, or at max 2-level dip. Probably after you have 1 or 3 levels of synth summoner.
Yeah I cannot find anything that is about punching people that isn't just a Fighter or Barbarian with a 2-hander sword that is any good. There's some rogue stuff that is nice, there's an alchemist archetype that gets sneak attack and would make a murderously effective hand to hand combatant with the mutagen and the buffs. But nothing based around jumping on things and punching people except synth summoner. Yeah. That's my advice.
Synthesist Summoner 20.
Optional; Monk 1.
Optional; Evangelist.
Best Answer
Don't make all of the bosses suddenly immune, but you can certainly add in strong henchmen with immunities. If I were a big baddie in your campaign (and had a pulse) then I'd be recruiting bodyguards who didn't care about con damage. Especially if the PCs aren't shy about their strategy. Even if so, some divination here and there wouldn't be game-breaking.
Depending on what ways your PCs are delivering con damage, you might try shifting the battlefield around. Is it the melee battlers? The next bad guy is an archer with some serious high terrain advantage. Magic? This would-be tyrant invested in some spell resistant cloaks.
Basically, the answer boils down to this: Your party is relying heavily on one strategy. Villains who don't prepare for that strategy die, and the rest thrive. Ever heard the phrase "there will always be somebody stronger?" Add in "survival of the fittest," too.