Improve performance for RandomIsomorphismTest to skip non-isomorphic checks for records with equal coderecs

[limakzi]

Currently, RandomIsomorphismTest may attempt isomorphism checks between records that already have equal coderecs, which is unnecessary and impacts performance. Update the procedure so that lists of records with equal coderecs are not redundantly tested for isomorphism, improving both speed and resource efficiency.

[fingolfin]

@limakzi was that meant to be an AI prompt? It doesn't quite read like an issue...

Anyway, in practice these function are not called with inputs that have identical codes.

[limakzi]

@fingolfin The input (first argument) for this function is a list of coderecs.

The output is a subset of those coderecs, filtered to remove any that are isomorphic to others in the list. My typical workflow involves generating a variety of pc-groups and then filtering them to retain only those that are likely non-isomorphic.
Groups with equal coderecs are isomorphic without involving the probabiblity test.

I think converting the input list to set is good optimization.