sync validate.yml with master

[geekosaur]

Too many backports in the wrong order = mayhem

Template B: This PR does not modify behaviour or interface

E.g. the PR only touches documentation or tests, does refactorings, etc.

Include the following checklist in your PR:

• Patches conform to the coding conventions.
• Is this a PR that fixes CI? If so, it will need to be backported to older cabal release branches (ask maintainers for directions).

[geekosaur]

I think I need to patch this up a bit

[geekosaur]

Fun. Windows had an unexpected pass due apparently to autotools (or possibly %TMP% being set). Reverted those until the relevant PR is backported (if ever).

[geekosaur]

Does anyone familiar with Macs know why ~all the Mac tests are dying with

*** Exception: /Users/runner/work/cabal/cabal/dist-newstyle-validate-ghc-9.10.1/build/x86_64-osx/ghc-9.10.1/cabal-testsuite-3/build/setup/setup: createProcess: execve: does not exist (No such file or directory)

? It's also happening in the 3.12 backport, but (obviously) not on master, and there's no obvious cause in the diff aside from the switch back to macos-latest, which we've been successfully running on master for around a week with no other code changes. I've purged the cache to verify that it's not something like Rosetta2 not working properly on the AArch64 runners (that is, that the precompiled setup isn't Intel instead of AArch64).

https://github.com/haskell/cabal/actions/runs/11309950048/job/31454413507?pr=10459 is a more complete example of how it's failing.

[geekosaur]

cabal-tests hates AArch64 for some reason. Presumably one of @jasagredo's PRs fixes it, since it works on master, but I'm not going to fight with it right now.