Re: [GHC] #7585: Core lint failure when optimizing coercions in branched axioms