[CI] Split default G01 deadline checks

[jjangg96]

Closes #64
Ref #1

Summary

Split only the default untagged G01 race suite into two static, disjoint 45-second processes: an exact TestBaselineStatisticsPresenceAndEligibility partition over ./..., followed by an unfiltered ./... remainder with only that exact name skipped. Keeping ./... package discovery in both partitions runs any same-named test in another package once in the first partition, while the unfiltered remainder preserves ordinary tests, Example Output, and fuzz seeds; G02 default checks and all existing tagged groups, exclusions, Go 1.26.8 pin, race, count=1, and failure propagation remain unchanged.

TDD evidence

Red before the script change:

$ GOTOOLCHAIN=go1.26.8 go test -count=1 ./scripts -run '^TestToolingDefaultG01PartitionsRun$' -v\n--- FAIL: TestToolingDefaultG01PartitionsRun\n... default G01 partition invocation ... ran 0 times; wrapper log: ... test -race -count=1 -timeout=45s ./...\nFAIL\n```\n\nExit status was 1 because the generated executable heavy/remainder/other-package/same-name/Example Output/Fuzz seed fixture saw only the old unpartitioned G01 invocation.\n\nGreen after the minimal split:\n\n```console\n$ GOTOOLCHAIN=go1.26.8 go test -count=1 ./scripts -run '^TestToolingDefaultG01PartitionsRun$' -v\n--- PASS: TestToolingDefaultG01PartitionsRun (26.02s)\nok .../scripts\n```\n\nThe positive fixture proves each generated witness runs exactly once; each generated failure is run independently and propagates a non-zero result.\n\n## Verification\n\n- `GOTOOLCHAIN=go1.26.8 go test -count=1 ./scripts -v` — PASS, 84.400s.\n- `GOTOOLCHAIN=go1.26.8 go test -race -count=1 ./scripts -v` — PASS, 84.787s.\n- `GOTOOLCHAIN=go1.26.8 go -C experiments/g01-scaleset test -count=1 -timeout=45s -run '^TestBaselineStatisticsPresenceAndEligibility$' ./...` — PASS, 12.08s wall.\n- `GOTOOLCHAIN=go1.26.8 go -C experiments/g01-scaleset test -count=1 -timeout=45s -skip '^TestBaselineStatisticsPresenceAndEligibility$' ./...` — PASS, 23.30s wall.\n- `GOTOOLCHAIN=go1.26.8 go -C experiments/g01-scaleset test -list . ./...` — PASS; 195 top-level tests: root 13, livecanary 106 (heavy exactly once), liveworker 76; no source examples/fuzz targets.\n- `GOTOOLCHAIN=go1.26.8 make fmt-check` — PASS.\n- `env GOTOOLCHAIN=go1.26.8 GO=go bash scripts/check-offline-experiments.sh` — PASS, 318.31s; G01 race groups were heavy 18.128s, remainder 33.626s/11.622s by package, and unchanged tagged groups 76.315s/41.074s/56.961s/43.951s; G02 passed.\n- `env GOTOOLCHAIN=go1.26.8 make check` — PASS, 472.95s; toolchain, format, build, vet, normal/race root tests, fuzz smoke, dependency/license checks, offline gate, and govulncheck all passed.\n- `git diff --check` and `bash -n scripts/check-offline-experiments.sh` — PASS.\n\n## Safety, limitations and rollback\n\nNo live endpoint, App, credential, runner, workflow, Docker/Lima, Keychain, launchd or service operation was used; no runtime files were changed. Local results do not establish hosted Linux reliability; independent Luna/max review, hosted CI, and exact-current-head Codex review remain required before merge.\n\nRollback is source-only: close the PR before merge or revert commit `69a3117a33f0c2b43210dc1bd5a60fef1db9961f`; no runtime state or runner cleanup is created.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-08T09:03:25.442809Z	69a3117	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[jjangg96]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Can't wait for the next one!

Reviewed commit: 69a3117a33

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[jjangg96]

Final gate: independent Luna (gpt-5.6-luna / max) read-only review APPROVE at 69a3117; no actionable findings. Independently reproduced old-script regression exit 1; generated six positive/failure witnesses passed normally (25.646s) and with race (26.945s). Actual race heavy/remainder partitions passed (13.053s/26.003s livecanary), G02 race passed (33.982s), tagged regression passed (49.230s), vets, bash syntax, gofmt and diff checks passed. Public CI run 34207430903 succeeded for this head. Codex completed clean at 2026-09-08T09:03:24Z naming 69a3117; wrapper all reports no findings, including historical findings. This offline CI-only change does not satisfy G01/G02 live gates or authorize live operations. Rollback: revert this focused change via a reviewed PR; no runtime migration exists. Conditional merge will target only this reviewed SHA.