feat(pytest_plugin): Add XDG persistent cache and fixture optimization

[tony]

Summary

• Add XDG-based persistent cache for pytest fixtures (~/.cache/libvcs/pytest/)
• Add RepoFixtureResult with cache hit/miss tracking
• Add atomic_init for race-free initialization with pytest-xdist
• Include FileLock module for atomic operations

Test plan

• Unit tests for FileLock
• Fixture performance benchmarks
• Cache hit/miss tracking verification

[codecov]

Codecov Report

❌ Patch coverage is 64.57086% with 355 lines in your changes missing coverage. Please review.
✅ Project coverage is 59.02%. Comparing base (d9391c3) to head (ed6812b).
⚠️ Report is 176 commits behind head on master.

Files with missing lines	Patch %	Lines
tests/test_fixture_performance.py	25.42%	176 Missing ⚠️
src/libvcs/_internal/file_lock.py	62.13%	72 Missing and 6 partials ⚠️
src/libvcs/pytest_plugin.py	51.18%	58 Missing and 4 partials ⚠️
conftest.py	50.66%	36 Missing and 1 partial ⚠️
tests/_internal/test_file_lock.py	99.44%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #502      +/-   ##
==========================================
+ Coverage   58.23%   59.02%   +0.78%     
==========================================
  Files          40       43       +3     
  Lines        6489     7456     +967     
  Branches     1098     1157      +59     
==========================================
+ Hits         3779     4401     +622     
- Misses       2177     2509     +332     
- Partials      533      546      +13     

☔ View full report in Codecov by Harness.
📢 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.

[tony]

Automated review finding. Recorded here so it is not lost. Nothing has been pushed to this branch and no fix is applied — this is a note, not a change request.

Vendored file_lock.py carries the stale-lock bug from #501

This PR vendors _internal/file_lock.py verbatim from #501, including its stale-lock handling, which decides staleness from mtime age alone and never checks whether the PID recorded in the lock is still running.

A slow-but-healthy holder therefore has its lock deleted out from under it, and two processes enter the guarded section together. Full detail and the suggested os.kill(pid, 0) liveness check are on #501; whichever lands first should carry the fix so the other does not reintroduce it.