Support definitely has a lot of strengths in its favor. You get access to, as far as I've seen, extremely rare and powerful spells like Haste, AoE debuffs, and Ember Burst, that to my knowledge can't be learned ingame except for from lucky attribute rolls. Most importantly, as a support you're never *truly* just a support (aside from the earlygame) because of the existence of items. Even if you don't have a single true heal, attack, or status clear in your repertoire you actually do have some of the strongest in the game in the form of items. It also synergizes naturally with being a summoner as your team won't suddenly lose a key feature while you're spending turns desummoning/summoning. (This is what I personally have the hardest time with regarding draining melee and Evade Step builds, which effectively stop functioning in their role once you stop to manage summons.)
Though frankly I don't use attack items enough, especially when I get them early, so that may be one of the problems I've been having with 1 1/2 consistency. However it does raise the question of whether aggregating chance (finding items, getting enough early EXP, getting zombies, getting lucky in the fight itself, or not running into Jiangshi first/at all on 1/2) is sufficient, or if that situation should be relatively reliably survivable as a rule. In the case of Mind Mask, I think Direct Gaze will already do the matchup an enormous amount of good since unlike zombies they ARE mind vulnerable. You're probably right in that Mind Blast would be too strong, that's an endgame-tier ability all on it's own. I haven't personally used it though so I don't have a firm grasp of just how high that stun rate is when you have +30% cunning, but I have been on the receiving end.
Really, when it comes to 1 1/2 it's level 1 jiangshi packs that pose the biggest threats. West, Zombies, Firecats, Ghouls, and even Chindis can all be relatively reliably managed using their capture mechanics no matter what loadout you're playing, and Slimes aren't much of a threat to begin with. What sets Jiangshi apart is their high damage stat layout, their chase and burst potential via Pounce, and the huge regeneration which serves as a DPT check to be able to kill them, and makes spending turns rescuing summons and healing become counterproductive. (Also the capture quest that makes them, if anything, even harder to fight.) If anything I'd say they're the true "boss fight" of the floor, rather than West and the zombies which you can relatively reliably just run away from. For that same reason they should be handled with a lot of care, as that kind of challenge is what roguelikes should strive for in the first place.
Like you said, it's something I'll need to keep challenging in order to better grasp my opinion on. There might even be a solid answer in the game already and I just haven't found it yet.
The level 2 thing is a bit weird though. The thing with filling the first 2 floors with weak demons is not just that the EXP is low, but also your recruitment options are also very weak. Notably with a full level 1 team the only demon that offers a good offense and defense in a single package is Wisp. Wisp also happens to be completely awesome, I never leave home without him.
You do raise a good point on the resists. I got it in my head that the resists are a balance point, but it's also true that their relevancy changes with the type prevalence within the game. (Then again the same goes for the balance of pretty much everything in the game.) For that reason it might be moot to worry about balance when you're in an active content/system development stage, I guess baseline playability is the only thing that currently matters, and the game is certainly winnable judging from the scoreboards. I almost feel like I've been thinking about the wrong things this entire time.
As an aside, I also might have a disproportionate phobia of Light weakness simply because I don't know the exact numbers of Hand of Light and thus have no idea if it might be capable of killing me or my demons at any given moment. Judging from the (rather amusing) tendency of my chindis to cast Snuff Out on 75% HP allies that have Guilt just to get the heal out of it, it's probably 0% chance for a much larger proportion than I think.
Not sure which code you're referring to, if you mean I'm going off-topic then I apologize, when I get into a brainstorm-y mode like this I have a hard time reigning back random thoughts. On the note of Pounce, I don't mind the idea of gap closers, I just feel like it (even compared to other gap closers) does an awful lot of things. Pounce basically teleports, attacks, and stuns in a single turn. Compare to Haunt which is just a teleport and a psuedo-stun, Bull Rush which to my knowledge doesn't ignore obstacles, and Leap Attack which is a rare lategame ability. Pounce is also more common and usually appears earlier than any other gap closer.
It's theoretically balanced by being a cooled ability, but the thing with cooldowns is that they disproportionately affect the player; your numbers are limited while the dungeon can always just throw more demons, with independent cooldowns, at you. When you're fighting a huge pack of Zars for instance, something (sometimes multiple things) can be getting pounced and panicked every single turn. (This is, in my opinion, one of the big reasons why Turdak's Sanctum is arguably the most dangerous floor in the entire game. The other being huge quantities of Autoantipathy lizards.)
I guess it's a question of what kind of role you want a move to serve. Right now pretty much every charge is extremely offensively potent, especially with numbers: it becomes stronger with larger numbers because of elements like chain stunning and cooldown cycling, and jumping over things makes it even better for ignoring the front row and dismantling whatever back row supports the player has set up for the encounter (and comparatively this is less beneficial for the player as you usually don't want to jump into the midst of large groups of enemies). If for instance you wanted a charge move that predominantly rewards the player for going melee, you can replace Stun with something that has diminishing returns when used in larger numbers. IE: A defensive or evasion self-buff, as defense buffs inherently are more valuable the more things you have attacking you. A non-stacking AoE debuff, excellent in the case of one user charging a crowd but less effective when you have a many users charging a few.
One might even say that extremely powerful offensive tools will inherently favor the dungeon, while powerful defensive tools favor the player. The reason for this is because the player's objective is just to survive to the win condition, while the objective of the dungeon is to kill the player.