fix: use more reduction when computing parent types (#7764)

This PR adds in more normalization for the routine that computes a
parent type. Some mathlib adaptations are the result of not reducing the
type parameters.
1 file changed