Allow containers-0.7

[Bodigrim]

https://hackage.haskell.org/package/containers-0.7 has been release last September, but in order to bump it in GHC submodule we need Cabal / Cabal-syntax HEAD to allow it. Should be a matter of a pure version bump to land into master, I do not foresee any incompatibilities.

[ulysses4ever]

Would be great to try the method of testing new bounds in CI described in https://github.com/haskell/cabal/blob/master/CONTRIBUTING.md#hackage-revisions

[Bodigrim]

@ulysses4ever I triggered https://github.com/haskell/cabal/actions/runs/7482718227 with allow-newer: *:containers and constraints: containers>=0.7.

[Bodigrim]

The errors in CI run are inconsequential: they boil down to a fact that the test suite expects containers unit id to be of form containers-<VERSION> (as it holds for non-overridden boot packages), but if we override it with --constraints 'containers>=0.7', the unit id has form containers-<VERSION>-<HASH>.

This can be solved (e. g., by using a non-reinstallable package instead of containers), but the easiest plan is to do nothing: unless the build plan is forced to choose non-boot version of containers, there will be no failures.

Could someone please take over? GHC 9.10 fork is looming and it's important to allow containers-0.7 before it happens.

(I'm not asking for a Hackage revision of existing packages, a Git commit would be enough at this stage)

[ulysses4ever]

@Bodigrim thanks a lot for your time and for trying the new workflow mode. The manual mode is still a pilot since we just added it. So we shouldn't hold anything important its hostage.

I'll ask on the Matrix if anyone is up for taking it over. If not, I hopefully will look into it after the ECOOP artifact deadline (23th).

[mpickering]

I think the issue with containers-0.7 not getting normalised is actually separate from it having a hash. It will be fixed by #9635 I expect, it's to do with the environment package database leaking into the testsuite.