`63c` and `63e` can never pass: run-by-name has no process serving the agent's workers

**Files:** `examples/agents/63c_run_by_name.py`, `examples/agents/63e_run_monitoring.py`, `examples/agents/63b_serve.py`, `examples/agents/63d_serve_from_package.py`

**Symptom:** Both hang indefinitely (300s timeout, no output) on every server type.

**Cause:** `runtime.run("doc_assistant", ...)` runs an agent **by name**, so the client never sees the agent object and cannot start workers for its tools. The tool tasks are scheduled with nothing polling them:
```
task 'search_docs' SCHEDULED in doc_assistant, queue last polled 6005s ago # 63c
task 'health_check' SCHEDULED in monitoring, queue last polled 600s ago # 63e
```
The examples meant to provide those workers cannot: `63b_serve.py` and `63d_serve_from_package.py` have every `runtime.serve(...)` call **commented out** — their executable bodies are a single `runtime.run(...)`. So no shipped example serves workers for a named agent.

**Fix:** Either restore a working `serve()` in `63b`/`63d`, or document that `63c`/`63e` require a separate serving process and show the command. Any fix should make the pairing runnable end to end.

**Verify:** With the serving process running, `python examples/agents/63c_run_by_name.py` completes instead of hanging.

**Related:** `run_all_examples.py`'s `DAEMON_SKIP` describes `63b`/`63d` as "serve() worker daemon (blocks forever)". They are not daemons — both complete in ~3.4s. That skip entry is stale; `63c`'s ("needs 63b_serve.py workers running in another terminal") is accurate but unachievable as shipped.

Start with examples/agents/63b_serve.py, 63d_serve_from_package.py, 63c_run_by_name.py, and 63e_run_monitoring.py, then check run_all_examples.py's DAEMON_SKIP entries. Choose either a working serve() pairing or documented separate serving commands, and run 63c with workers available to confirm it completes instead of hanging.

索引モデルが issue の本文から書いたものです。