ci: make e2e tests resilient to CI resource contention

E2E specs spin up Docker containers in parallel and intermittently time
out waiting for UI elements under shared-runner load. Limit parallel
workers to 2 and retry flaky failures twice so the container job can
complete and publish the PR preview image.

package.json

@@ -49,6 +49,7 @@
     "test": "cross-env NODE_ENV=development CI=true vitest run --project \"!e2e\" --retry 2 --coverage.enabled ",
     "test:e2e": "cross-env NODE_ENV=development CI=true vitest e2e --retry 2 --maxWorkers 2",
     "test:workshop": "sh apps/workshop/test.sh",
+    "test:workshop-live-provider": "node apps/workshop/tests/openrouter-live-smoke.mjs",
     "test:workshop-image": "sh apps/workshop/test-image.sh",
     "test:ui": "cross-env NODE_ENV=development CI=true vitest --project \"!e2e\" --ui --coverage.enabled",
     "typecheck": "turbo typecheck",