feat: add Experiment lifecycle recovery

[Cedric Vidal (cedricvidal)]

Summary

Adds recoverable Experiment deletion and restore as the fifth stacked Experiment successor PR.

• Persists lifecycle identity, phase, CAS version, exact chunk-backed request/report target snapshots, progress, blockers, and deletion provenance.
• Freezes Experiment mutation and publication before cascading; queue claims and worker final writes reject deleted or lifecycle-active records.
• Restores only children owned by the exact operation, preserves independently deleted records, and rolls to a fresh execution when the restored request/run vector drifted.
• Repairs interrupted operations from dispatcher polling, isolates per-Experiment repair failures, and rotates blocked work fairly.
• Adds project-scoped REST, CLI, Portal, Storybook, migration, OpenAPI, and architecture/website documentation surfaces.

Stacked on #1516.

Demos

Before

N/A — Experiment lifecycle management did not previously exist.

After

Lifecycle list visibility, deletion progress, retained deleted state, child-first restore, and reconciled active state:

experiment-lifecycle.mp4

Deletion progress:

Restore progress:

Testing

• pnpm test — passed 294 files / 3,528 tests before final review fixes.
• pnpm test — passed 293 files / 3,529 tests after final review fixes; one unrelated Windows PowerShell test timed out.
• pnpm exec vitest run scripts/build-windows-worker.test.ts — passed 8/8 on immediate isolated rerun.
• Consolidated lifecycle coverage — passed 130/130 tests before final review; focused lifecycle/coordinator/migration regressions passed 18/18 after review fixes.
• Production builds passed for shared, telemetry, db-migrations, API, CLI, Portal, readout-handler, report-generator, Storybook, and documentation website.
• OpenAPI generation/snapshot, Storybook play coverage, website tests/build, and git diff --check passed.
• Compose smoke attempted once with the isolated worktree project; blocked before startup by host Podman ENOSPC at /var/tmp/libpod_builder…. No prune, delete, or retry was performed.
• Walkthrough take 2 validated as H.264/yuv420p, 1440×900, 25 fps, faststart; full decode and representative encoded-frame inspection passed.

Documentation and compatibility

• Updated docs/architecture/experiments.md and docs/architecture/app-design.md.
• Updated the website Experiment guide and regenerated the committed OpenAPI document.
• Migration 036 adds Cosmos-compatible lifecycle polling and deletion-provenance indexes.
• Existing active Experiment behavior remains compatible; deleted records remain hidden unless includeDeleted / Show deleted is explicit.

Checklist

• If Portal features changed, keep CLI capabilities in sync.
• If Portal components changed, update their Storybook stories.
• If database changes require a migration, include up() / down() and keep it CosmosDB-compatible.
• If dependencies changed, update the lockfile and regenerate NOTICE / NOTICE-REVIEW.txt with pnpm notice as needed. (N/A — no dependency changes.)
• Video showing the behavior before the suggested change (N/A — new capability.)
• Video showing the behavior after the suggested change

[github-actions]

Test Results (Node.js 22)

test: Run #283

Tests 📝	Passed ✅	Failed ❌	Skipped ⏭️	Pending ⏳	Other ❓	Flaky 🍂	Duration ⏱️
3530	3526	4	0	0	0	0	2m1s

❌ Some tests failed!

Name	Failure Message
❌ ExperimentStore CAS contracts > plans chunked-to-chunked growth and chunked-to-inline shrink independently	Test timed out in 5000ms. If this is a long-running test, pass a timeout value as the last argument or configure it globally with "testTimeout".
❌ ExperimentDetail > starts an asynchronous deletion from the confirmation dialog	Test timed out in 5000ms. If this is a long-running test, pass a timeout value as the last argument or configure it globally with "testTimeout".
❌ ExperimentDetail > shows deleted lifecycle progress and restores the Experiment	Test timed out in 5000ms. If this is a long-running test, pass a timeout value as the last argument or configure it globally with "testTimeout".
❌ ExperimentDetail > removes selected current members in one revision-guarded request	Test timed out in 5000ms. If this is a long-running test, pass a timeout value as the last argument or configure it globally with "testTimeout".

Github Test Reporter

[github-actions]

Test Results (Node.js 22)

test: Run #294

Tests 📝	Passed ✅	Failed ❌	Skipped ⏭️	Pending ⏳	Other ❓	Flaky 🍂	Duration ⏱️
3535	3535	0	0	0	0	0	1m17s

🎉 All tests passed!

Github Test Reporter

[github-actions]

Test Results (Node.js 22)

test: Run #299

Tests 📝	Passed ✅	Failed ❌	Skipped ⏭️	Pending ⏳	Other ❓	Flaky 🍂	Duration ⏱️
3535	3534	1	0	0	0	0	1m49s

❌ Some tests failed!

Name	Failure Message
❌ ExperimentStore CAS contracts > plans chunked-to-chunked growth and chunked-to-inline shrink independently	Test timed out in 5000ms. If this is a long-running test, pass a timeout value as the last argument or configure it globally with "testTimeout".

Github Test Reporter