[thekevinbot]

Refs #1481

./** in a git repo: each walk thread now matches a .gitignore through its own compiled copy.

• Why. Gitignore::matched borrows its match buffer from a pool. Only the pool's first caller gets it without a lock; the walk's other threads contend on the rest. In a microbenchmark of 1.6M paths, matching on 32 threads took 30.7ms through one shared Gitignore and 7.7ms through one copy per thread.
• How. A new Frame (gitignore_frame.rs) wraps the Gitignore and keeps a lazily built copy per thread slot, compiled from the same file by the same GitignoreBuilder calls. Matching semantics stay with the ignore crate. If a copy no longer compiles, matching falls back to the shared Gitignore.

Measured on a 3.2M-item corpus built by glob_speed_gitignore_test.build_tree (1.84M visible files), hyperfine, 40 runs, host load ~17, 32 cores. Medians:

Query	main	this PR	fd
SELECT count(*) FROM './**' (walk only)	0.156s, 2.61s user	0.152s, 2.26s user
SELECT path FROM './**'	0.295s	0.273s	0.306s

The walk uses 13% less CPU. The wall-clock gain is small on an idle 32-core host and larger under load.

This PR alone does not close #1481. Most of the gap is in the output path, which PR #1513 fixes. In glob_speed_gitignore_test:

• main: 1.20x vs fd (load 27)
• PR Speed: ./**/*.md at or under fd #1513's head: 0.81x vs fd (load 14)

Once #1513 merges, the test's bar can be pinned to FASTEST.

Changelog / Migrations

• Changelog fragment added under packages/rust/changelog.d/ for each changed package
• Migration fragment with the same slug added under packages/rust/migrations.d/ for each changed package (every subsection but Summary _None._)

E2E Verification

• Ran e2e suites locally for every affected SDK
• Python SDK e2e: pass (16 passed, impacted subset)
• TypeScript SDK e2e: pass (18 passed)
• Rust core e2e: pass (cargo test --workspace --features cli)
• packages/python/e2e-attestations/<branch>.json receipt written
• packages/ts/e2e-attestations/<branch>.json receipt written
• Command(s) run: testing-conventions e2e attest 'uv run python -m pytest tests/e2e/glob_speed_gitignore_test.py tests/e2e/glob_speed_matchers_test.py tests/e2e/glob_speed_recursive_test.py tests/e2e/query_contract_test.py -x -q -s' (in packages/python, after maturin develop --release and cargo build --release -p dirsql --features cli), testing-conventions e2e attest 'pnpm test:e2e' (in packages/ts), cargo test --workspace --features cli
• Result summary: all pass. SELECT path FROM './**' output is byte-identical to main (md5 over 1.84M rows).