[chenBright]

What problem does this PR solve?

Issue Number: resolve #1120

Problem Summary:

When event dispatchers occupy all bthread workers in epoll_wait or kevent,
other bthreads cannot run. This can stall request processing and server shutdown,
particularly when event_dispatcher_num >= num_threads.

What is changed and the side effects?

Changed:

• Add bthread_add_reserved_concurrency_by_tag(num, tag) to increment reserved concurrency.
• Reserve an extra worker for each event dispatcher and internal epoll bthread. Reserved workers are
  created eagerly, including when lazy growth is enabled.
• Track normal and reserved concurrency separately. Keep TaskControl::concurrency() as their total.
• Use independent per-tag limits of 1024 normal workers and 512 reserved workers, allowing reservations
  even when normal concurrency reaches its limit.

Side effects:

• Performance effects:

• Breaking backward compatibility:

---

Check List:

• Please make sure your changes are compilable.
• When providing us with a new feature, it is best to add related tests.
• Please follow Contributor Covenant Code of Conduct.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

[Copilot]

🟡 Changes recommended

Three unresolved moderate findings remain: the shared epoll thread can pin workers to the first caller's bthread tag, dispatcher reservation failures now abort at startup via CHECK_EQ, and a test timeout can leave a thread holding dangling stack references.

3 open issues

• Avoid crashing when dispatcher reservation fails · New
• Document or fix epoll thread tag placement · New
• Prevent dangling references when stop_and_join times out · New

_{Review effort: Lite}

[wasphin]

LGTM