Associative, commutative, etc "traits" for access patterns

[gussmith23]

A lot of Glenside's rewrites focus on implementing simple things, i.e. "bubbling" an access pattern through another access pattern (aka using composition commutativity to reorder the access patterns). I would love it if we could implement a way to automatically derive these types of rewrites. At first, we could use unproven claims that access patterns have specific traits (i.e. it commutes with another access pattern). Perhaps in the future, we can check these claims, or even automatically infer claims.