fix(telegram): clear send_path_degraded immediately on successful reconnect

[annguyenNous]

Fixes #35205

Problem

After a transient network interruption, the Telegram gateway successfully reconnects polling but _send_path_degraded remains True for up to 60 seconds (until the heartbeat probe runs in _verify_polling_after_reconnect()), blocking all outbound send() calls during that window.

Root Cause

In _handle_polling_network_error(), the reconnect ladder at line ~964 resets _polling_network_error_count = 0 but does NOT clear _send_path_degraded. The flag is only cleared inside _verify_polling_after_reconnect() (line ~1024) which runs after a 60-second HEARTBEAT_PROBE_DELAY.

Fix

Clear _send_path_degraded immediately when start_polling() succeeds in the reconnect ladder, restoring the send path without waiting for the deferred probe. The probe still runs as a safety net and will re-set the flag if the connection is actually wedged.

Changes

• gateway/platforms/telegram.py: +9 lines — clear _send_path_degraded after successful reconnect with info log

[alt-glitch]

Duplicate of #35286 — both fix #35205 (send_path_degraded stays true after reconnect). #35286 includes test updates; recommend closing this in favor of that PR.

[alpindiay]

Automated Triage: Duplicate Detected

This PR's changes are a strict subset of #35286 (same send_path_degraded fix, same file). #35286 includes the same fix plus test coverage. Recommend closing this PR in favor of #35286.

[teknium1]

Automated hermes-sweeper review: this reconnect fix is already present on current main.

Evidence:

• The Telegram adapter now lives at plugins/platforms/telegram/adapter.py after the gateway platform plugin migration.
• In _handle_polling_network_error(), a successful start_polling() path clears self._send_path_degraded = False immediately at plugins/platforms/telegram/adapter.py:1776, before the deferred heartbeat probe.
• Regression coverage exists in tests/gateway/test_telegram_send_path_health.py:106, where the test disables _verify_polling_after_reconnect() and verifies the flag clears plus send() works immediately.
• The implementation landed in 2ecb6f7fe60f6a240d632bbd51bcbb25ad22c161 (fix(telegram): clear send_path_degraded on successful reconnect (#35205) (#54076)).

Thanks for the focused fix here; the prior duplicate discussion was accurate, and the requested behavior is now covered on main.