Re: [GHC] #12176: Failure of bidirectional type inference at the kind level