test: cover empty-mapping rebase regression

[svarlamov]

Fixes #1476

Summary

• Add a TestRepo regression for an empty-mapping rebase followed by a real rebase to the same origin/main target.
• Prove the first rebase actually drops the already-upstream equivalent commit before exercising the later rebase.
• Verify the amended AI commit keeps its authorship note, exact line attribution, and 2 AI / 0 unknown stats after the second rebase.

The live reflog-segment resolver implicated by #1476 was removed when #1542 replaced it with trace2/ref-cursor processing. The old proposed RebaseSkipped fix in #1478 now targets dead code, so this PR locks in the current behavior without reviving that architecture.

Test coverage

• Historical red check: the regression fails at the final note assertion on the issue-era parent commit ef8bc3157.
• task test TEST_FILTER=test_empty_mapping_rebase_does_not_block_later_rebase_to_same_target
• task lint
• task fmt
• task test was run twice; the new regression passed both times. Each run ended on a different unrelated daemon-sync flake, and both flaky tests passed immediately in isolation:
  • test_author_config_partial_overrides_fall_back_to_git_committer_identity
  • test_nested_subrepo_mixed_edits_mock_ai

---

[svarlamov]

• test: cover empty-mapping rebase regression #1909 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.