#7416 fix(cli): report backend readiness after startup

Summary

Fixes #7252.

The dev CLI could report a running backend after its worker failed to import the app. Granian readiness now follows successful ASGI lifespan startup.

What changed

• Report unexpected worker exits while preserving recovery after file changes.
• Report frontend readiness separately from backend startup.
• Accept both frontend message formats in reflex-bench.
• Cover import failures, lifespan failures, delayed startup and repeated recovery with real workers.

Verification

• Full unit suite: 11,196 passed, 155 skipped; 79.04% coverage.
• Repository Ruff checks, formatting and Pyright passed. Changed-file pre-commit hooks passed.
• Real backend-only and combined Reflex CLI runs recovered from import failures. The benchmark driver waited for recovery; frontend HTTP and backend /ping returned 200. All test processes were stopped.

Deployment notes

Tools parsing Granian dev logs should recognize Frontend running at: for the frontend URL and wait for Backend running at: for backend startup. Production and legacy Uvicorn messages are unchanged.