fix: drain general httpx connection pool on Telegram send pool timeout

[jackdaw-talks]

Summary

When Telegram's httpx connection pool for general requests exhausts due to
stuck connections, only the polling pool is drained. This adds drain for send pool.

See commit for full details.

[liuhao1024]

✅ Verified — httpx connection pool exhaustion fix

Reviewed the diff for Telegram send/edit path reliability under pool exhaustion.

• New method _drain_general_connections() at gateway/platforms/telegram.py:945: Correctly targets only _request[1] (general/send pool), leaving _request[0] (getUpdates polling) untouched — polling continues uninterrupted during drain
• Pool lifecycle: shutdown() → initialize() sequence is correct for httpx pool reset
• Integration points: drain is called in 3 error paths — send retry (send() line 2093), send error return (send() line 2162), and edit_message error (edit_message() line 2323). All wrapped in try/except for non-fatal handling
• Internal API access (_request[1]) acknowledged with # noqa: SLF001

The fix addresses a real production scenario: half-closed connections from proxy interruptions (e.g., sing-box) gradually exhausting the pool until all sends fail with Pool timeout. Thorough coverage of all error paths. No issues found.

[teknium1]

Automated hermes-sweeper review: this older PR's fix is already implemented on current main.

Evidence:

• plugins/platforms/telegram/adapter.py:1670 defines _drain_general_connections_after_pool_timeout(), which resets PTB's general Bot API request pool after a confirmed send-side pool timeout.
• plugins/platforms/telegram/adapter.py:3185 calls that drain helper from the send retry path when _looks_like_pool_timeout(send_err) is true.
• tests/gateway/test_telegram_thread_fallback.py:1343 covers the exact behavior: the general request pool is shut down and initialized, while the polling request pool is left untouched.
• The implementation is present on current HEAD via 7e2ca7f68da63a29e1695b5e24901cc57e32f2ac (fix(telegram): reset send pool after pool timeouts). The later fix(telegram): reset send pool after pool timeouts #53622/fix(telegram): refresh general request pool after pool timeout #53624 discussion also records that this fix was salvaged through fix(telegram): drain general send pool on pool timeout before retry (#53524) #54121.

Thanks for the diagnosis and patch. The behavior this PR targeted is now covered on main, so closing this as implemented there.