repo: add head_commit_info property for studio live experiments monorepo support

[mattseddon]

• ❗ 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. 🙏

---

Follow up to #10283. Related to https://github.com/iterative/studio/issues/8848.

This PR adds a head_commit_info property to the Repo object. This property provides the information required to correctly display live experiments generated from monorepos in Studio. The information is sent via dvc_studio_client.post_live_metrics as dvc_experiment_parent_data.

Notes:

• This data collection doesn't fit intuitively/easily anywhere within the DVC ecosystem
• The property will be used by both dvc and dvclive
• I considered adding the collection to dvc-studio-client but that would involve making dvc-studio-client aware of scmrepo and/or GitCommits.

Is there a better place to put this functionality?

[codecov]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (44b78b8) 90.43% compared to head (b1ad23d) 90.43%.

Files	Patch %	Lines
dvc/repo/__init__.py	90.90%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #10284   +/-   ##
=======================================
  Coverage   90.43%   90.43%           
=======================================
  Files         494      494           
  Lines       37811    37841   +30     
  Branches     5468     5478   +10     
=======================================
+ Hits        34193    34222   +29     
  Misses       2985     2985           
- Partials      633      634    +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[skshetry]

Could you share an example how this will be used?

[mattseddon]

Could you share an example how this will be used?

@skshetry see iterative/dvclive#774 & #10285

[skshetry]

@mattseddon, Repo has a large API surface. So, we don't usually add any API directly to it as possible.
We have split up the parts of Repo into smaller APIs, like repo.scm for example, that houses all git-related procedures and functions.

repo.scm has APIs that gives you all required information, as you have used in the PR already. What you have in head_commit_info is a very specific serialization of GitCommit for Studio, so I don't think it should a Repo property.

Either this should be handled in appropriate places, in dvclive and in dvc separately, or at best, it should be a utility function somewhere in dvc.scm/dvc.utils.studio.

I'd prefer the former, as they are independent clients and can have independent implementations, but I have nothing against the latter.

I considered adding the collection to dvc-studio-client but that would involve making dvc-studio-client aware of scmrepo and/or GitCommits.

In medium/long-term, I'd like to move away from dvc-studio-client. It tries to be independent of dvc, but it has to try to re-implement parts of dvc.
dvclive already depends on dvc and I see no point in maintaining dvc-studio-client at the moment. It just creates issues like this.

[mattseddon]

I'll move everything around.

[mattseddon]

dvclive already depends on dvc and I see no point in maintaining dvc-studio-client at the moment. It just creates issues like this.

@skshetry is the suggestion here to move dvc-studio-client into dvc? I can take that if that's what you want to do.