Freeze hackage Index state

It turns out that head.hackage CI has started failing due to changes in Hackage. We should really freeze the state of the Hackage index for GHC pipelines to ensure that we don't start seeing spurious failures such as this.

There is, however, the question of whether all head.hackage pipelines should use a frozen index or just those triggered from ghc/ghc>. In principle we could do the latter and use a nightly pipeline in the head.hackage repo to catch breakage originating from Hackage.