>Paul Erlich has mentioned that he's been messing with a >"well-temperament" variation in his 22TET work
This is right in my paper (have you read it?). 12-tone well-temperaments give certain heptatonic keys with good overall 5-limit harmony and other keys with better 3-limit harmony but bad ratios of 5. Similarly, the 22-tone well-temperament in my paper gives certain decatonic keys with good overall 7-limit harmony, and other decatonic keys which approach Indian tuning, having better 5-limit harmony but bad ratios of 7.
>but I doubt he considers >it significantly de-22-ated. (-ized?)
Basically, that's true, although the 5-limit keys in 12-tone well-temperaments sound like they could come from 31, and the 5-limit keys in 22-tone well-temperaments sound like they could come from 34.