Re: [GHC] #16296: OccurAnal should not break the linter assumptions