Assuming you can't get a theme, here's my suggestion for a damage-focused crossbow sniper:
====== Created Using Wizards of the Coast D&D Character Builder ======
Drow, Thief, Dread Fang
Background: Occupation - Criminal (+2 to Stealth)
FINAL ABILITY SCORES
Str 9, Con 12, Dex 24, Int 11, Wis 20, Cha 11.
STARTING ABILITY SCORES
Str 8, Con 11, Dex 18, Int 10, Wis 14, Cha 10.
AC: 29 Fort: 21 Reflex: 30 Will: 25
HP: 89 Surges: 7 Surge Value: 22
Stealth +23, Thievery +19, Streetwise +12, Dungeoneering +17, Athletics +11, Acrobatics +19, Nature +17, Perception +17
Arcana +7, Bluff +7, Diplomacy +7, Endurance +8, Heal +12, History +7, Insight +12, Intimidate +9, Religion +7
Level 1: Ruthless Hunter
Level 2: Primal Sharpshooter
Level 4: Two-Fisted Shooter
Level 6: Backstabber
Level 8: Crossbow Expertise
Level 10: Weapon Focus (Crossbow) (retrained to Lasting Frost at Level 11)
Level 11: Primal Eye
Level 12: Wintertouched
Level 14: Silvery Glow
Lolthtouched: Cloud of Darkness
Primal Sharpshooter: Grappling Spirits
Thief utility 1: Tactical Trick
Thief utility 1: Escape Artist's Trick
Thief utility 2: Fleeting Ghost
Thief utility 4: Sneak's Trick
Thief utility 6: Chameleon
Thief utility 7: Ambush Trick
Thief utility 10: Counter-Step
Frost Hand Crossbow +3, Feytouched Drowmesh +3, Bracers of the Perfect Shot (paragon tier), Assassin's Cloak +3, Eagle Eye Goggles (heroic tier), Frost Hand Crossbow +2, Potion of Healing (heroic tier)
====== Copy to Clipboard and Press the Import Button on the Summary Tab ======
I calculate DPR with to-hit chances, so my numbers will be much lower than yours.
Ranged basic attack: +23 v. AC, 1d8+26 cold damage
Combat advantage from 2 sources: wintertouched/lasting frost with a 1-(28-23)/20=75% chance and hiding based on invisibility or sneak's trick (any source of concealment will do)
When you start your turn hidden (for extra juice there, take persistent tail as your level 10 utility), you deal 4d8 as your sneak attack damage.
Tohit: (1-(28-(23+2))/20) = 85% hit rate, with backstabber means you generally won't miss.
DPR: .8*(4.5+26+4.5*4+5)+.05*(8+26+8*4+5+3.5*3+.85*(4.5+26+4.5*4+5)=49.14 DPR, all encounter long every encounter. Backstab to be used to turn misses into hits and you're a walking damage machine. AP of course gets you a second verse same as the first, so that's 100 damage on the first round. at 136 average HP for level 14, the slowed enemy (grappling spirits) gets a round to croak for help before you finish them in the second round. No dailies in the class which makes for an incredibly long-enduring character.
Free reroll on stealth checks from your neckslot with a standard action invis from your armor gives you all the stealth you really need. Especially with sneak's trick allowing remarkable stealth in places that would not otherwise be considered venues for sneaking. Average non-trained perception is +10 at 14, so a +23 with a reroll and an II to become re-hidden should provide for sufficient stealth.
For a nova-based assassin... I'll look at the executioner with big, sad, eyes... and go to the Ranger. Their damage simply cannot be beat. I'd personally choose a wind-rider, but that's because I find pure-damage frostcheese excruciatingly boring. The darkstrider gets you your hidden damage boost. Spending feats on skills is usually contraindicated. See here for a goodenough archer that just spams the damage.
If you choose to not go with thief (cunning sneak is slightly less good than sneak's trick, depending) or ranger, then executioner isn't a bad choice. You're trading damage for flavour, but it's very nice flavour. If you MC warlock and take shadow dancer then it's very nice, very very very very mobile flavour. Warlocks also make good assassins MC into assassin :)
Cunning sneak is just not that worth it in actual play. While it helps in theoretical "solo stealth" missions, it's hard being a meat&potatoes assassin. Bravo's a trap. If you're going PH1 rogue/ranger, focus on minor action attacks, multi-attacks, and frostcheese. They'll get you the most damage with the least tradeoffs. Darkstrider becomes better there, because 3+wis damage on multi-attacks very very quickly becomes stupid.
Don't forget to calculate to-hit in your damage potential.
Good luck. Feedback will be incorporated to refine requirements and my suggestions.
It sounds like you have a very acceptable fire-focused intention. I suspect one of the things that is complicating matters is the fact that:
The Mage Wizard can select 2 Encounter powers per level, so that offers some variety. It helps that the DM lets us waive the restrictions on the number of powers we can use per day. I think that's why he makes the monsters so strong in each encounter, so they don't get wiped out easily.
Therefore, much of the normal optimisation advise, which assumes that you're holding to the normal rules starts melting away as the vicious circle of buff and counterbuff begins. (I faced this problem in a "by the rules" game when the DM reacted to the party's increasing optimisation by ramping up monsters, which caused us to optimise more, which...)
From a pragmatic perspective, save ends effects suck. While much of the game is well modelled, there is precious little balance to save ends effects, and the pendulum swings back and forth: standard monsters have little to no defense, but elites and solos become effectively immune as the game design progressed through the monster manuals. It takes a very deft touch in monster creation (if you're creating monsters "from scratch" to respect player agency in the inflicting of status while simply not going "nope!" to either them automatically winning or to them automatically being ignored.)
I, personally, have always enjoyed the more controlly-type controllers, and so my wizards, druids, psions, and invokers have focused on debuffing and forced movement. So long as you rely on effects that are more difficult to shed (either being end of next turn or encounter long) then you can focus on being to reliably land them, rather than inflicting sufficient debuffs to the monster's saving throws (that'll only be countered by the next monster) to maintain the debuff. The same thing is true in the other direction. I've played paladins who granted +9 to saving throws by smiling. This led to the DM completely foregoing the use of save-ends effects until the DM and I agreed to voluntarily limit that feat to a +5 bonus.
My recommendations are:
Nothing is as powerful alone compared to a party that is designed to work together.
Stop focusing on solo optmisation. It's a trap. Instead, try to make sure the party is designed to work together to achieve your desired requirements. Everyone will have more fun, and you're unlikely to bear the brunt of your DM's nerfing alone.
Have a side conversation with your DM: Explore what debuffs he's comfortable with.
Boundary setting is important. If you have a chat over coffee as to what he considers reasonable, you won't find the powers nerfed in the middle of a game. Set up, describe, and agree upon expectations for your character's capabilities such that he knows what to expect (such as to provide you maximum Fun) with the minimum of unpleasant surprises. As 4e is very much combat-as-sport, the joy is in the execution of plans within a chosen narrative (yes, story matters, to provide a need and justification for mechanics) than it is finding unusual solutions to the DM's prepared set-piece battles (many other systems are far far better at simulation).
It's very hard to alter characters in midstream without a retcon. Be honest and do a proper retcon, don't just knudge.
A character is the combination of her parts and their interactions, not just the parts alone. If you're changing a character's rasion d'etre, be honest about it, and change the character completely to fit your new requirements.
This answer is a conservative justification of the low values presented in DPR King 2.0:
After some deliberation, I agree with these assessments. Given that enemies will be avoiding bunching up save for the opportunity to flank, the heavy discount on unfriendly AoEs seems to correspond to how often there will be a combat where:
1. There are enough enemies
Given that the number of enemies in a fight optimally decreases with respect to time (you don't want to leave everyone on 1 hp until the last round) blasts and bursts have a decreased time where they can be effective.
Given that a non-trivial fraction of the fights are versus solo, elite, or small numbers of people, fulfilling the parent condition is difficult.
2. They are close enough without a party member in the way
Enemies will maneuver to avoid AoEs, just like party members. The most likely people to catch are the meleers flanking in the dogpile/scrum/charlie foxtrot, whatever you want to call the ball where all the melee happens. Targeting party unfriendly powers is hard, and almost certainly cannot happen every round.
Now, with that said, there are all kinds of ways to virtually expand the size of blasts and bursts, and anything that just adds "one square" onto a blast or burst can be said to upgrade its size for purposes of this chart.