fix: write file cache entries atomically to avoid read races

[7487]

Description

Fixes #6010

FileCache.put opened the cache entry with "wb", which truncates the file in place before the gzip/pickle write completes. A concurrent reader (e.g. another pytest-xdist worker sharing the cache directory) can open the file inside that window and hit pickle.load EOF ("Ran out of input"), which is what intermittently broke tests/lsp/test_reference_macro_find_all.py::test_multi_repo_macro_references on Windows CI.

This change writes the payload to a temp file in the same cache directory and os.replaces it onto the entry path, so readers only ever observe complete entries. Details:

• The temp file is prefixed with the cache version so the stale-file cleanup in FileCache.__init__ doesn't unlink it mid-write (fresh atime keeps it alive; orphaned temp files from a crashed process still get cleaned up by the atime threshold).
• The store is best-effort: on Windows os.replace fails with a sharing violation if a reader still has the target open. In that case we log a warning and keep the existing intact entry rather than raising (mirroring how get handles unreadable entries).

I didn't add the optional get retry from the issue — get already swallows unpickling errors and get_or_load falls back to the loader, and with atomic writes a partial read can no longer happen.

Test Plan

• Added test_file_cache_put_is_atomic, which verifies that a failed os.replace leaves the existing entry untouched and cleans up the temp file.
• pytest tests/utils/test_cache.py passes.
• Stress-checked with 2 readers looping get against a writer looping put on the same entry: on main this produced ~2900 failed reads (partial file -> unpickle EOF); with this change, 0.

Checklist

• I have run make style and fixed any issues (ruff check, ruff format and mypy are clean on the touched files)
• I have added tests for my changes (if applicable)
• All existing tests pass (ran tests/utils; remaining fast-test failures in my env are missing optional engine deps, unrelated to this change)
• My commits are signed off (git commit -s) per the DCO

[cmgoffena13]

Your try wraps a lot here. I would keep the try/finally to unlink, but add in a try/except for the os.replace specifically, like this:

try:
    with...
         ....
    try:
         os.replace
     except OSError as ex:
          warning...
finally:
    try:
    ....

[cmgoffena13]

I would just cut this down to: "Windows os.replace fails if a concurrent reader still has the target file open."

[cmgoffena13]

@7487 -- Would you please comment on the issue saying you're working on it. I can't assign an issue unless you've commented. Thanks!