Run E2E against a Dockerized SUT

[omermorad]

What changed

• move the sample SUT to a root-level Docker service
• start, health-check, log, and tear down Compose independently in every E2E shard
• classify SUT lifecycle failures as infrastructure failures without masking test failures
• select Playwright projects and tag-based labels independently before sharding
• support all/any label matching and fail closed when zero tests execute
• expose project and label filters for manual and reusable workflow calls

Verification

• Docker image built and reached healthy
• full Playwright suite: 4 passed
• API project plus @smoke: 2 passed
• single @regression test across two shards: merged 1/1 passed
• startup, missing evidence, and secondary cleanup failure simulations remained fail closed
• TypeScript, JavaScript syntax, YAML parsing, and diff checks passed

[github-actions]

E2E test report

Signal	Result
Gate	✅ Passed
Tests	3 passed · 0 failed · 0 skipped · 3 total
Profile	pull-request
Projects	api, chromium
Labels	@smoke · match any
SUT	Docker Compose, isolated per shard
Runner image	playwright-e2e-runner:e2e-4b90af156d5626fc1b8d9839c530dd4662190e0d2ca6d2b1bf95600d79bb2e6d · sha256:b8387478d0c1d9e0f479af55f3f9d70d3f3b404ad6ec016a25f876fb8dd3c8a3 · e2e-runner-image · cache hit
Report image	playwright-e2e-report:36383735978-1 · sha256:00b803054c3df1b90f28ea3d7225b52537237cc53d34880f94f7b993137f1e93 · e2e-report-image
Results	Open workflow run · e2e-results

Updated from run 36383735978.

[github-actions]

E2E test report

Gate	✅ passed
Tests	3 passed · 0 failed · 0 skipped · 3 total
Profile	pull-request
Projects	api, chromium
Tags	@smoke (any)
Runner	playwright-e2e-runner:e2e-4b90af156d5626fc1b8d9839c530dd4662190e0d2ca6d2b1bf95600d79bb2e6d (reused)
Results	e2e-results
Report image	playwright-e2e-report:36431128971-1 · e2e-report-image
Run	Open workflow run