fix(mutation): move the repo size ratchet where the lane cannot reach it

[thymikee]

Repairs the mutation baseline on main, as asked on #1964.

The defect

Stryker runs the suite from a sandbox copy under .tmp/stryker/. A test that asserts about the repository checkout itself — its files on disk, or its git history — therefore reads a repository that does not exist.

test-file-size-ratchet.test.ts is such a gate, and it fails there for two independent reasons:

1. disableTypeChecks (Stryker's default) prepends // @ts-nocheck to every copied {test,src,lib}/**/*.ts, so all 26 pinned files read one line longer than they are.
2. The sandbox has no origin/main, so the gate's history-backed half cannot resolve its merge-base.

Fixing either leaves the other — I confirmed that the hard way, by fixing (1) first and watching CI fail on (2). Its own .tmp skip entry cannot help either: that is matched relative to REPO_ROOT, which inside the sandbox is the sandbox.

The fix

Move the gate to scripts/__tests__/ and include it explicitly in unit-core — the address this repo already uses for maintained gates that are not src tests (the help-conformance, xctest-selection and package-closure gates all live there).

KERNEL_TEST_FILE_RE admits only root/package src tests, and its own comment already names scripts/__tests__ as unreachable by construction. So the gate leaves every mutation lane by virtue of where it lives: no classifier to recognise it, nothing to keep in sync, and no way for a future lane to pull it back in without deliberately widening that pattern.

Nothing about the gate's behavior changes. REPO_ROOT is ../.. from either address, and TEST_ROOTS already included scripts, so it measures exactly the same tree — 4/4 tests pass unmoved.

What this replaces, and why

The previous revision added a repoIntrospectionTestFiles scanner that classified tests by source text — a single-quoted walk-files import, or the string origin/main. That was the wrong boundary, as reviewed: a behavioral test could match it and be silently excluded, while an equivalent repo gate written with double quotes, a different walker, or a different base ref would be missed. The scanner, its test, and its justifying comment are all deleted; scripts/mutation/test-scope.ts is now byte-identical to main.

The one assertion added instead pins the invariant this fix depends on: isKernelTestFile accepts root/package src tests and rejects scripts/__tests__. That is the whole invariant, not a sample of it — widening the pattern would silently pull the gate back into every lane, and now fails loudly at that edit instead of confusingly during a dry run.

Verification

pnpm mutation:run --modules kernel-errors at this commit, .tmp cleared first:

• scope 804 → 803 test files — exactly this gate
• dry run clean; lane envelope pass at stage complete
• score 74.8% (187 killed / 63 survived / 250, 0 timeouts) — unchanged from the pre-existing baseline, as expected: a file-length-and-history gate can never kill a kernel mutant
• pnpm mutation:test 39/39 · pnpm check:layering 181/181 · typecheck, lint, format clean

stryker.config.json is untouched, so the config content hash the report and lane envelope carry is unchanged and scores stay comparable across this commit.

Whose fault this was

Nobody's, and in particular not #1964's. You and that PR's author independently measured 804 kernel-related test files on both the exact base and its head, with a zero-line diff between the lists. Before #1969 rewired the module graph, nothing pulled this gate into a lane's scope, so the trap stayed hidden.

[github-actions]

Size Report

Metric	Base	Current	Diff
JS raw	2.37 MB	2.37 MB	0 B
JS gzip	794.8 kB	794.8 kB	0 B
npm tarball	916.5 kB	916.5 kB	0 B
npm unpacked	3.18 MB	3.18 MB	0 B

npm unpacked components

Component	Base	Current	Diff
JS / dist source	2.51 MB	2.51 MB	0 B
Apple runner source/project	564.2 kB	564.2 kB	0 B
macOS helper source	54.5 kB	54.5 kB	0 B
Android helper artifacts	0 B	0 B	0 B
Other package files	44.5 kB	44.5 kB	0 B

Startup median (7 runs, lower is better):

Scenario	Base	Current	Diff
CLI --version	18.7 ms	18.7 ms	+0.0 ms
CLI --help	51.3 ms	52.4 ms	+1.1 ms

Top changed chunks: no changes in the largest emitted chunks.

Top changed packed files

No changed packed files.

[thymikee]

The CI diagnosis is valid and every exact-head lane is green, but repoIntrospectionTestFiles is the wrong boundary. It classifies tests by source-text sniffing for one single-quoted walk-files import or origin/main: a behavioral test can match and be silently excluded, while an equivalent repo gate using double quotes, another walker, or another base ref is missed. The test proves only one current behavioral file is absent, not the claimed invariant.

Use the repository’s existing structural owner instead: move test-file-size-ratchet.test.ts under scripts/__tests__/ and explicitly include it in unit-core in vitest.config.ts. Mutation scope already admits only root/package src tests through isKernelTestFile, so the repo gate then becomes unreachable by construction and this scanner, speculative test, and paragraph-long justification can be deleted.