fs: stop stat()ing every entry in recursive readdir

[codebytere]

Makes fs.readdir(), fs.readdirSync() and fsPromises.readdir() with { recursive: true } use the entry types the binding already returns instead of a stat() per entry, and build relative names by prefix instead of path.join() + path.relative() per entry. Results, their order and the symlink-following behavior are unchanged.

benchmark/fs/bench-readdir.js / bench-readdirSync.js (gain a recursive option), recursive=true, 30 runs:

	lib/ (473 entries)	test/parallel/ (~4 000 entries)
readdirSync	+272 %	+676 %
readdirSync, withFileTypes	+124 %	+402 %
readdir	+93 %	+582 %
readdir, withFileTypes	+47 %	+344 %

The recursive variants asked the binding for names only and then called internalModuleStat() on every entry to find the directories to descend into; with withFileTypes they built the Dirents and still stat()ed every entry that was not already a directory (to follow symbolic links, #52663). On a 13 500-entry tree that is 13 500 extra stat() calls, and the sync variant was ~4.5x slower than a hand-written readdirSync(dir, { withFileTypes: true }) walk.

Now the binding is always asked for types, directories are descended into directly, and only symbolic links and entries of unknown type (the ones that could still lead to a directory) are stat()ed. Output was compared entry-for-entry against main on test/ and on a tree with file/directory/broken symlinks and a FIFO. The known_issues test for Buffer paths (#58892) now asserts success: the ERR_INVALID_ARG_TYPE reaches the callback instead of being thrown from the completion handler, so with a bare mustCall() it would have started passing.

Tests: test-fs-readdir-recursive*, test-fs-readdir-types*, test-fs-opendir and fs.cp tests pass.

---

Disclosure: the code, measurements and this description were written by Claude Code, directed and reviewed by @codebytere.

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (5858205).
⚠️ Report is 38 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff            @@
##             main   #65487    +/-   ##
========================================
  Coverage   90.12%   90.13%            
========================================
  Files         752      751     -1     
  Lines      252315   252458   +143     
  Branches    47444    47509    +65     
========================================
+ Hits       227395   227545   +150     
- Misses      16217    16220     +3     
+ Partials     8703     8693    -10     

Files with missing lines	Coverage Δ
lib/fs.js	98.38% <100.00%> (-0.01%)	⬇️
lib/internal/fs/promises.js	92.41% <100.00%> (-0.11%)	⬇️

... and 58 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/76246/