[RPG] the volume of one Alchemist’s Fire

A spell is resolved - and the target is determined - when the spell comes into effect, not when spellcasting is "starting".

So in the example the sorcerer could still cast the spell, just another target would be determined if they are no longer aware of where the wizard is. The only way to stop the spell being cast (Spell resistance, saving throws, etc are another matter) would be to counterspell or disrupt (Concentration checks) the spell itself - removing the (intended) target before the spell is complete simply means that the caster (the sorcerer in this case) is free to choose a new target.

Ie. Even though the sorcerer had taunted the wizard the spell target hadn't yet been determined.

This is usually most relevant when a spell takes more than a simple action to cast (such as sleep, a one round to cast spell) but when reactionary effects get in the way spellcasters can still choose their target if their original one vanishes before their spell can take effect.

It's detailed in the Casting Time section of the Magic rules of Pathfinder, but the most important line of it is:

You make all pertinent decisions about a spell (range, target, area, effect, version, and so forth) when the spell comes into effect.

So, in the fraction of a moment where the wizard teleports away the spell's effect has not been determined - the wizard hasn't made a saving throw. So the sorcerer is still free to target anyone else they want. If there are no relevant targets of course (say, charm animal and all the animals vanish into the ether on a joyride) then obviously the spell will fail through lack of targets.

Encouraging inventiveness is easy. You've already done it, in fact: You simply have to reward it by allowing it to (occasionally) succeed.

Avoiding breaking the game is also easy: You simply disallow anything that lets players to bypass the obstacles presented by the game with less effort than it would take to tackle those obstacles in a more traditional way.

Obviously, these two goals are hard to reconcile, but I can suggest a few things that might help.

First, let your players have the benefit of their invention, but remember to account for the costs: In your example, your alchemist has developed a devastating weapon, but one which eats through very expensive ammunition at a disturbingly rapid rate. Once the disparity between party member wealth levels grows a bit, he might start to worry. Also, given that it is a devastating weapon, even if it is expensive, ask yourself why other people don't use it. I imagine it would take just one orc with the Improved Sunder feat to make your player realise that there are pitfalls involved in waving acid vials over his own head.

Second, remember to include at least a few situations in which the invention does not work, to keep your players on their toes (and encourage future inventions): There's no end of creatures immune to acid and fire, so let them crop up occasionally - and your alchemist would be in strife if away from civilisation for an extended period.

Finally, don't be afraid to condtradict previous rulings if you can come up with a good reason for it. Yes, you should avoid it most of the time, but inventions are a special case, since by definition they're trying something new and unknown: Invention is inextricably paired with discovery. Be tactful, and start by saying "There's something I didn't think of that's kind of relevant to that idea you came up with," but be firm, and try and provide a reasonable in-game justification for it. "Your acid net? The ropes that make it up only have about two hit points, so it should be disintegrating every time you use it. Also, how have you been protecting yourself from the splash?"