fix: distinguish error cases in list_refs

[effigies]

list_refs() now returns descriptive strings instead of None for error cases: "repo-not-found" when the remote repository doesn't exist, "repo-empty" when the repository exists but has no refs, and "command-failed" for other git ls-remote failures.

The stderr check scans for "fatal:" and "not found" to robustly identify missing repositories even when the remote sends additional diagnostic lines before the fatal message.

Breaking change, but will be shortly followed up by patches on the dashboard.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.85%. Comparing base (68cf92e) to head (f50d7ae).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main       #4   +/-   ##
=======================================
  Coverage   97.85%   97.85%           
=======================================
  Files          17       17           
  Lines        1260     1261    +1     
=======================================
+ Hits         1233     1234    +1     
  Misses         27       27           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.