hermes - 💡(How to fix) Fix Telegram TimedOut exceptions crash entire gateway (affects all profiles)

Bug Description

Unhandled telegram.error.TimedOut exceptions crash the entire Hermes gateway process, affecting all agent profiles (default/Henry, jake, sven, alfred).

Steps to Reproduce

1. Telegram API experiences slow response (network latency, API degradation)
2. Gateway attempts to send a message via Telegram
3. telegram.error.TimedOut is raised
4. Exception is unhandled → gateway process dies
5. systemd restarts gateway after 5 seconds (Restart=always)

Evidence from Logs

telegram.error.TimedOut: Timed out
...
hermes-gateway.service: Scheduled restart job, restart counter is at 11

From journalctl --user -u hermes-gateway:

• Crash trace shows telegram/request/_httpxrequest.py raising TimedOut
• Gateway dies immediately on unhandled exception
• All profiles affected (jake logs show multiple Timeout crashes)

Impact

• All profiles crash: Henry, Jake, Sven, Alfred all experience gateway restarts
• Conversation interruption: Active turns are killed mid-response (confirmed 4+ times in one session)
• User experience: Messages appear to send but responses are lost
• Restart loop: Restart=always + unhandled exceptions = crash/restart cycle

Environment

• OS: Linux Mint (6.17.0-29-generic)
• Hermes Agent: latest main
• Telegram bot: polling mode
• systemd service with Restart=always

Expected Behavior

Gateway should:

1. Catch telegram.error.TimedOut
2. Implement retry/backoff logic
3. Log warning but not crash
4. Maintain conversation state across temporary API issues

Suggested Fix

In gateway/platforms/telegram.py, wrap message sends in try/except:

try:
    msg = await self._bot.send_message(...)
except telegram.error.TimedOut:
    logger.warning("Telegram API timeout, retrying...")
    # Implement retry with backoff
    # Don't crash the gateway

Workaround

Currently none — all profiles affected, systemd restarts gateway after 5s but active conversation is lost.