Allow running CI from rawhide images

[mtrmac]

We will, temporarily, want to run from rawhide images that include the rust-podman-sequoia RPM, as well as Fedora 43 images.

We have only one skopeo_cidev though, and we don't want to build a rawhide variant; so, allow consuming it from rawhide.

See individual commit messages for details.

A prerequisite for Sequoia containers/image#2876 , see https://github.com/containers/image/pull/2876/checks?check_run_id=48518809749 .

@lsm5 PTAL.

[lsm5]

@mtrmac LGTM if we need this happening soon. @cevich PTAL.

BTW, with the testing-farm jobs already in place and assuming #2640 will get merged (hopefully soon), would we need to worry about having this? IIRC, one reason to still have this would be to run tests locally without polluting your environment, but maybe the same could be achieved by running a toolbox container (?)

[cevich]

Another use-case could be get_ci_vm.sh but I'm doubting that gets used here much. I can't tell, but was a temporary commit used to confirm this logic actually works with rawhide?

[lsm5]

LGTM

@mtrmac please merge whenever convenient.

[lsm5]

Ignore f43 testing-farm job. Compose isn't ready yet AFAICT. The testing-farm people will hopefully have a process change in place for f44 so that issue shouldn't occur from the next release forward.

[mtrmac]

BTW, with the testing-farm jobs already in place and assuming #2640 will get merged (hopefully soon), would we need to worry about having this?

We will need CI coverage of the Sequoia build configuration, probably in situations consistent with how we build distribution RPMs.

IIRC, one reason to still have this would be to run tests locally without polluting your environment

I’m not positively aware of anyone doing this — I rely on GitHub-connected CI ~exclusively. So I don’t have a strong opinion on whether / how that should be possible.

[mtrmac]

I can't tell, but was a temporary commit used to confirm this logic actually works with rawhide?

c/image #2876 was building against these images ; I’m trying to split the Skopeo PR into the CI-prerequisite + new-feature parts , and I didn’t actually test that the split is correct — I failed in a previous iteration.

Hopefully this will get it right; if not, I’ll do a proper isolated test c/image PR.