perf(git): single porcelain status via git-dir state detection

[iagogfe]

Summary

• Avoid a second plain git status on the compact path (hot path for rtk git status)
• Recover detached HEAD and in-progress state (rebase/merge/cherry-pick/revert/bisect/am/sparse) from git-dir marker files and porcelain unmerged codes
• Measured ~30% faster wall time for rtk git status (eliminating the dual-spawn was the main cost)

Test plan

• cargo fmt --all --check && cargo clippy --all-targets && cargo test
• Manual testing: rtk git status output inspected
• Unit tests for prose parsers kept under #[cfg(test)]; git module tests pass
• Perf (local, Linux x86_64, release build, hyperfine --shell=none, 80 runs):
  • before (dual git status): ~15.8 ms mean
  • after (single porcelain + git-dir state): ~10.9 ms mean
  • ~1.45× faster (~5 ms absolute) on this machine — not a universal guarantee

Important: All PRs must target the develop branch (not master).
See CONTRIBUTING.md for details.

[CLAassistant]

All committers have signed the CLA.

[tapheret2]

Review: perf(git): single porcelain status via git-dir state detection

Files: src/cmds/git/git.rs
Size: +211 / -24

Notes

• Static pass on the patch: no obvious blockers from the diff alone.

Thanks @iagogfe — independent review on the patch.