Sorry, bit autistic and all. Don't mean to be confrontational except in the sense that I'd like to provoke thinking about the problem in a different way.
I program for a living. You are asking for a system (code) without actually describing what you want (specs). What I was showing was that different specs would produce radically different products while all meeting the overly loose requirements of, "Create a system for arbitrating the outcome of a magic pool"
I don't usually formally track loot, but one way to look at this is that they are taking magic items out of the pool and hense increasing their wealth level. So, if the benefits are largely positive, you have to count this against their expected wealth by level. If you do that, then the system is as balanced (or unbalanced) as D&D's wealth by level system and CR/EL system itself. Nothing else is needed in theory. I mean, "Can use overland flight 1/day" is more or less giving the character an item that can do that for practically all purposes.
Now, if your system is truly Gygaxian, then it basically balances itself in the sense at least that it balances reward with risk. And yours seems to offer lots of risk with the rewards, so as long as your are fine with some players getting screwed and others getting rewarded, then you are balanced.
You are running all of these as if they are quirks, but they are all special cases of 'looks like a freak'. In my game, I usually run 'freak' disadvantages as either a penalty on diplomacy checks with first acquaintances (once people get to know you, they stop judging you by your looks), or by asserting that everyone reacts to you one step more negatively (ei, normally indifferent characters are unfriendly, and normally unfriendly ones are hostile). I tend to prefer the latter, though in some cases they stack.
I don't know the mechanics of most of the citations you make, but I'd be inclined to consider 'looks like a freak' a default drawback to many of them, and I'd be especially tempted to apply the drawback as an additional drawback whenever 2 or more physical mutations were picked up.
Shouldn't this be tied to character alignment? I'd probably run the above as, "If you are Chaotic, gain smite Law, otherwise suffer 3d8 anarchic damage." or something along those lines.
Also, if you are going the Gygaxian route, I find the idea of 'player's choice' which you use in some cases to be odd. In practice, that's probably a bad idea, in that negatives don't in fact balance with positives in those cases. Boosting your skill in something you do all the time is very difficult to balance against a negative in a skill you never use and rarely if ever would have to use because some skillful person will be nearby to do it for you.
I would consider that overly harsh. I would only apply such costs to being able to choose from the table. So, maybe one entry is, "Lose 1500 XP. Player gets to choose one benefit."
