Test hosted API cold-start recovery without hiding the wait

[Ayush-Kumar0207]

Problem

The public backend may cold-start on free infrastructure. The UI should communicate that state, recover automatically, and preserve local work instead of looking broken or spinning indefinitely.

What good looks like

• add a deterministic browser contract that delays the first health or workspace request long enough to enter the waking state;
• verify plain-language progress, a bounded retry policy, and a manual retry control;
• verify recovery when the API becomes healthy;
• verify local edits are not discarded during the wait;
• verify the terminal error state includes a useful next action;
• keep the test independent of the real production host.

Likely files

• client/tests/e2e/production-contracts.spec.ts
• the shared API readiness or status component
• focused unit tests for retry timing if needed

Do not shorten the production timeout merely to make the test faster; use routed responses or injected timing.