Knowing what to save your categorical mit for isn't "risk and reward". It just reduces the net complexity of optimizing one's sustain, since so much of the finer details (and maximum permissible incoming damage relative to tank base eHP at the encounter's tuning point) will be next to nothing compared to the simple mitigation types, which would then determine which (likely singular) CD schedule is permissible.
Rather, you're simply replacing...
- mitigation that can play a part in most mechanics/attacks and with therefore greater net potential for cognitive load (especially, if damage were increased to make tank swaps worthwhile without necessitating vuln swap or move-the-AoE-out gimmicks)
--with--
- mitigation that can meaningfully play a part only within certain attacks and frequency of action to see even the small degree of cognitive load already present.
By all means, see how much DRKs and GNBs enjoyed having their categorically-specific raid-mitigation button rendered useless to press in certain fights.
Chances are, tanks wouldn't get much more fun out of having their capacities arbitrarily reduced in a given fight if the portions of each mitigation type varied across tanks nor would they enjoy needing 3x the buttons to achieve the same agency they have now even if the tanks were homogenized to carry the same portions of each miti type (as to prevent certain tanks just being outright stronger/weaker in certain encounters).
No, I'd much prefer to keep mitigation mostly general and simply amp up our agency through increased frequency and interaction among mitigation tools (atop perhaps a fair bit more randomization and some added bankability). Simon Says provides far less overall engagement, let alone agency, than being more able to choose how much potential mitigation to invest over what windows of time.