feat: per-script smoke timings as a standing dataset (smoke_timings.json)

[Jammy2211]

Overview

The smoke runner's per-entry timings exist only as [PASS] <name> — <n>s lines in job logs, recovered by hand-scraping every time someone asks what is slow. Since the smoke-runner delegation (#260–#263) all ten workspace runners are thin shims over autohands/run_python.py, so recording per-script timings routinely is one PyAutoHands change, not ten repo sweeps. Phase 2 of the test-performance board arc (PyAutoLabs/PyAutoHeart#163); answers item 4 of the Mind's draft/research/ci/smoke_timing_and_profiling.md with "yes". Mind prompt: active/smoke_timings_dataset.md.

Plan

• The report machinery (result_collector.RunReport, already mandatory in the PR gate via --report-dir) additionally emits a consolidated smoke_timings.json: one entry per executed script/notebook — {entry, kind, status, seconds, cap_s, exit_code} — plus run metadata (project, env profile, python version).
• When $GITHUB_STEP_SUMMARY is set, append a compact per-entry timing table (slowest first) so every gate run's timings are one click away with no artifact download.
• PyAutoHeart's reusable smoke-tests.yml uploads the report dir as a run artifact (if: always(), if-no-files-found: ignore) — the dataset persists the artifact retention window for every gate run across all ten repos (separate small PR on PyAutoHeart).
• Timing comes from the runner's own measurement, never re-derived; TIMEOUT entries record the cap they hit; skips are never silently timed as 0.
• Existing report consumers (run_all.py, Heart test_run/script_timing) unaffected.

Detailed implementation plan

Affected Repositories

• PyAutoHands (primary)
• PyAutoHeart (one artifact-upload step in the reusable workflow, separate PR)

Branches

• PyAutoHands: claude/test-performance-dashboard-y3fdy7 (from main @ e510de1)
• PyAutoHeart: claude/smoke-timings-artifact-y3fdy7 (the phase-0 fix holds the main task branch there)

Implementation Steps

1. autohands/result_collector.py — RunReport gains per-entry timing capture (reusing the durations it already records for the per-script JSONs) and a write_timings() leg emitting smoke_timings.json alongside the existing report files; step-summary table writer guarded on GITHUB_STEP_SUMMARY.
2. autohands/run_python.py / run_notebook.py — no interface change; the report path emits the new file whenever --report-dir is passed.
3. PyAutoHeart/.github/workflows/smoke-tests.yml — upload-artifact step for the report dir.
4. Tests: entry shape incl. TIMEOUT carrying cap_s; skips absent/marked; step-summary table golden test; existing report consumers untouched (full suite green).

Original Prompt

Click to expand starting prompt

See PyAutoMind/draft/feature/pyautohands/smoke_timings_dataset.md (moved to active/ on issue) — full task text and acceptance criteria there.