Telegram channel crashes gateway with unhandled promise rejection on fetch failure

Description

The Telegram channel's long-polling causes the entire gateway to crash when a fetch request fails. The promise rejection from the polling fetch is not caught, triggering Node's unhandled rejection handler.

Environment

• OpenClaw version: 2026.1.29
• Node.js: v24.13.0
• OS: Windows 10

Reproduction

1. Enable Telegram channel
2. Wait for any network blip or timeout during Telegram API polling
3. Gateway crashes

Logs

[openclaw] Unhandled promise rejection: TypeError: fetch failed 
  at node:internal/deps/undici/undici:16416:13 
  at processTicksAndRejections (node:internal/process/task_queues:103:5)

Crash occurs ~15-60 seconds after gateway becomes idle, even with no agent runs happening. Only Telegram polling is active.

Analysis

• Crashes happen on a semi-regular interval matching Telegram's long-polling behavior
• Third crash in logs occurred with zero agent runs between gateway startup and crash — only Telegram was running
• telegram: autoSelectFamily=false (default-node22) in logs may be related (IPv4/IPv6 resolution)

Expected behavior

Telegram polling should catch fetch errors gracefully and retry, not crash the gateway.

Workaround

Disable Telegram channel:

{
  channels: { telegram: { enabled: false } },
  plugins: { entries: { telegram: { enabled: false } } }
}