perf: batch git exec calls for notes, rev-parse, status, and checkpoints

[svarlamov]

Summary

This PR implements performance optimizations to reduce git subprocess overhead by batching git exec calls. The primary bottleneck in git-ai is the number of individual git invocations, especially for operations like git notes show and git rev-parse that are called repeatedly in loops.

Changes:

• Fix Fix init command #1 & Initial test suit #2: Batch-fetch authorship notes in blame.rs and refs.rs using 2 git calls (1 cat-file --batch-check + 1 cat-file --batch) instead of N individual git notes show calls
• Fix PR Comment Demos #4: Combine resolve_commit + resolve_parent into single git rev-parse call in diff.rs
• Fix Cursor extension #5: Skip redundant get_staged_filenames() call in status() when no pathspecs provided (porcelain v2 output already contains staging info)
• Fix make git-ai init idempotent #7: Read checkpoints once instead of twice in get_all_tracked_files()
• Add batch_get_authorship_logs() helper function for reuse across codebase
• Make batch_read_blobs_with_oids() public for cross-module access

Expected impact: Significant reduction in subprocess overhead for blame operations (N commits → 2 git calls), diff operations (2 calls → 1), and status operations (2 calls → 1 when no pathspecs).

Updates since last revision

• Resolved merge conflict with main in src/commands/diff.rs: the resolve_two_revs function now uses exec_git_with_profile(&args, InternalGitProfile::General)? (combining the PR's ? error propagation + stdout parsing with main's migration from exec_git to exec_git_with_profile). This matches the pattern already used by resolve_commit in the same file.

Review & Testing Checklist for Human

⚠️ Risk level: YELLOW - Core git operations modified without end-to-end testing

• Test blame command with commits that have authorship notes, especially files with many unique commits. Verify AI/human attribution is correct and performance improves.
• Test diff command on edge cases: initial commits (no parent), symbolic refs (HEAD, branch names), commit ranges. Verify correct fallback to empty tree hash. Pay attention to the merge-conflict resolution in resolve_two_revs — confirm error propagation via ? behaves correctly when revisions can't be resolved.
• Test status command with staged files, both with and without pathspecs. Verify staged files are correctly identified without the explicit get_staged_filenames() pre-call.
• Verify schema version handling: batch_get_authorship_logs() correctly filters by AUTHORSHIP_LOG_VERSION, but the inline batch code in get_commits_with_notes_from_list() does not — it deserializes without a version check. Confirm this difference is intentional.

Notes

• Only cargo build and cargo clippy were run — no functional tests executed
• The status.rs change assumes porcelain v2 output correctly identifies staged files; this should be verified with real repos
• The diff.rs resolve_two_revs now propagates errors with ? instead of the old match on Ok/Err; initial commits fall back to the empty tree hash via the caller's match on resolve_two_revs
• In get_diff_json_filtered, the parent is resolved via resolve_two_revs(repo, &parent_rev, &to_commit) which redundantly re-resolves to_commit — minor inefficiency but not a bug

---

Link to Devin run: https://app.devin.ai/sessions/972f0c2bf6d94b17b663d3e7d52da147
Requested by: @svarlamov

---

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[git-ai-cloud-dev]

Stats powered by Git AI

🧠 you    ████████████████████  98%
🤖 ai     ░░░░░░░░░░░░░░░░░░░█  2%

More stats

• 1.4 lines generated for every 1 accepted
• 19 seconds waiting for AI
• Top model: claude::claude-sonnet-4-6 (62 accepted lines, 26 generated lines)

AI code tracked with git-ai

[git-ai-cloud]

No AI authorship found for these commits. Please install git-ai to start tracking AI generated code in your commits.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[devin-ai-integration]

✅ Devin Review: No Issues Found

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

View in Devin Review to see 5 additional findings.