Re: [GHC] #11203: Kind inference with SigTvs is wrong