Re: [GHC] #9196: Higher-rank constraint treated as type instead