exp show: show exp when checkpoint is running(#7329)

[karajan1001]

Fixes #7329

In a previous PR, we introduce a bug which use HEAD as the workspace
commit, while if there is a checkpoint is running, the HEAD is
actually changing all the time, we should use
refs/exps/exec/EXEC_BASELINE instead.

A test is needed for this PR.

1. replace the default HEAD in exp show with fix_exp_head.
2. modify a functional test to simulate the running conditions.

• ❗ I have followed the Contributing to DVC checklist.

• 📖 If this PR requires documentation updates, I have created a separate PR (or issue, at least) in dvc.org and linked it here.

Thank you for the contribution - we'll try to review it as soon as possible. 🙏

[pmrowla]

So the actual cause of this bug is that we don't handle fixing HEAD in iter_revs:

https://github.com/iterative/dvc/blob/e430693fc81b0ba27131756039e3191d0b93d83d/dvc/scm.py#L157

I think what we should really do for this is to make dvc.scm.resolve_rev() use fix_exp_head, and get rid of all of the other calls to fix_exp_head from elsewhere in DVC, otherwise we will probably keep running into this type of bug in the future.

and then we can just have one test for scm.resolve_rev() that does the same thing as:

https://github.com/iterative/dvc/blob/e430693fc81b0ba27131756039e3191d0b93d83d/tests/func/experiments/test_experiments.py#L625-L636

At some point what we really want is to make scm.resolve_rev() use fix_exp_head, and get rid of all of the other fix_exp_head calls everywhere else in DVC,

[karajan1001]

Currently, scmrepo is decoupled with this experiments related setting? Maybe it is better to put this logic in dvc/repo/experiments/utils.py instead of scmrepo. BTW, the import logic of dvc/repo/experiments/utils.py and dvc/scm.py is a little bit tricky, sometimes utils.py depends on something from scm.py sometimes the dependency comes from the opposite direction.

And besides this, I don't know if there are any conditions that we need to resolve the current temp local HEAD instead of the exp_head?

[pmrowla]

Currently, scmrepo is decoupled with this experiments related setting? Maybe it is better to put this logic in dvc/repo/experiments/utils.py instead of scmrepo. BTW, the import logic of dvc/repo/experiments/utils.py and dvc/scm.py is a little bit tricky, sometimes utils.py depends on something from scm.py sometimes the dependency comes from the opposite direction.

This does not belong in scmrepo for sure, dvc.scm is for scm/git behavior that is DVC specific (like this issue).

The main issue with keeping fix_exp_head separate from dvc.scm.resolve_rev is that this affects every DVC command (such asdvc diff), not just exp commands.

And besides this, I don't know if there are any conditions that we need to resolve the current temp local HEAD instead of the exp_head?

In the event that we need the local head, we can just use scm.get_ref("HEAD") directly.

[karajan1001]

OK dvc.scm.resolve_rev (dvc/scm.py::resolve_rev) is different from dvc.scm.resolve_rev (dvc("Repo").scm("Git")::resolve_rev) quite confusing.

I would suggest renaming it here to make them more clarified.