feat(ci): publish branchless performance reports

[zoeyrose]

Summary

• repair the trusted daily benchmark to run the exact complete client-benchmark matrix and preserve explicit early-failure diagnostics
• replace generated-branch persistence with a bounded, digest-validated, no-JavaScript Pages site and versioned JSON checkpoint
• retain detailed points separately from compact trend/state, enforce 90-point, cohort, file, and byte bounds, and physically prune aged files
• validate the last successful Pages deployment or the recorded final generated-data commit before deterministic projection
• isolate benchmark/projection, Pages deployment, and alert permissions; remove every report-path contents: write
• reconcile regression issues idempotently only after a successful deployment
• document interpretation, bootstrap, rerun/backfill, recovery, provider outage, and post-merge activation

Supports #311. The issue remains the manual post-merge close path because its 30-consecutive-point rollout criterion cannot be satisfied by merging code.

Governance companion: atrinik/github-settings#71

Coordinates

• Base: main at 1f98d430a03e6d9fbf6f27f4ca1356542d905d08
• Head: feat/benchmark-pages at aaf359d3aae4c41f2c5d89395b9f4410b1b7a62e
• Worktree: /workspaces/atrinik/workspace/worktrees/classic/issue-311-benchmark-pages
• Commits: 2279042 feat(ci): publish branchless performance reports; aaf359d fix(ci): harden performance checkpoint recovery
• Frozen bootstrap commit: bab40ecefefa5b6052d42eab6390c504b9482e81

Validation

• python3 -m unittest discover -s tools/tests -p 'test_*.py' — 233 passed
• python3 -m compileall -q tools/ci tools/tests
• python3 tools/verify_import_history.py — five imported histories verified
• actionlint .github/workflows/*.yml
• exact frozen-data bootstrap rehearsal — generation 1 retained historical run 31755454661 plus a fresh compatible projection
• git diff --check 1f98d430a03e6d9fbf6f27f4ca1356542d905d08..HEAD

Two complete whole-diff review rounds finished with zero known actionable findings. The focused tests cover byte-stable reconstruction, split projection, manifest/detailed-point/report binding, higher-attempt replacement, watermarks and global pruning, HTML escaping/no JavaScript, tamper and stale-file failure, synthetic alert regression/recovery, partial-operation retry deduplication, ambiguous issue ownership, exact matrix/trust restrictions, pinned Actions, and job permission isolation.

Verification and activation

Gameplay runtime is not applicable: this change owns CI evidence projection and static publication, not client/server behavior. Before merge, use the validation commands above.

After this PR and the governance companion merge, an organization owner follows the governance runbook to switch the exact Classic Pages site to the Actions source. Then dispatch Daily Classic client performance from main once with checkpoint_source=final-benchmark-data, verify the raw/checkpoint artifacts and deployed endpoints, and use checkpoint_source=pages for every later manual run. Do not delete or update benchmark-data; its final commit remains rollback evidence.

The prior site remains live on benchmark, predecessor, projection, validation, or deployment failure. Human rollout evidence—manual bootstrap, scheduled continuity, synthetic live alert transition, and at least 30 compatible daily points—remains tracked on #311 after merge.

[codecov]

Codecov Report

❌ Patch coverage is 59.72461% with 234 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
tools/ci/daily_performance_site.py	57.64%	139 Missing and 52 partials ⚠️
tools/ci/reconcile_performance_alerts.py	63.15%	25 Missing and 10 partials ⚠️
tools/ci/daily_performance_report.py	77.14%	4 Missing and 4 partials ⚠️

📢 Thoughts on this report? Let us know!

[zoeyrose]

Final whole-diff review is complete at aaf359d3aae4c41f2c5d89395b9f4410b1b7a62e.

• zero known actionable findings remain after two review/fix rounds
• all 233 local tooling tests, compile checks, import-history verification, actionlint, and diff checks passed
• the exact frozen benchmark-data bootstrap rehearsal produced a valid generation-1 site retaining historical run 31755454661 and a fresh point
• all latest-head required checks passed, including the Classic validation aggregator, native Windows security, CodeQL, and patch coverage
• rendered PR body, unchanged base, mergeability, comments, and review threads were rechecked; no unresolved threads exist

The Codecov informational comment was reviewed. Its required codecov/patch status is successful, and the new corruption, stale-state, identity, retry, pruning, and legacy-bootstrap boundaries have explicit regression coverage.

This PR is ready to merge. Issue #311 intentionally remains open for its post-merge activation and 30-compatible-point rollout evidence.