feat(listener): automatic database reconnection with bounded backoff

[MaybeSomeone-arc18]

Closes #786

Problem

When the listener's database connection drops (network blip, idle timeout, server restart), the connection is not re-established. Every subsequent query fails and the listener needs a manual restart to recover.

Fix

• listener/src/database/database.ts: the database layer now detects a dead connection and reconnects automatically with bounded exponential backoff, so transient drops recover on their own while a genuinely unavailable database fails fast instead of spinning.
• listener/src/utils/request-id.ts: carries the minimal docstring repair from fix: repair mangled docstring in request-id.ts breaking the build on main #846 so this branch builds and reviews independently (drops out on rebase once fix: repair mangled docstring in request-id.ts breaking the build on main #846 lands).

Tests

listener/src/__tests__/database-recovery.test.ts (new) exercises drop-and-reconnect recovery: queries succeed after a killed connection, backoff stays bounded under repeated failures, and in-flight reconnects do not stampede. Verified locally against the listener test suite.

[drips-wave]

@MaybeSomeone-arc18 Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits

[MaybeSomeone-arc18]

Hey, quick nudge - the wave closes tomorrow (Sep 30) 6:30 PM IST. This one's ready and tested locally: the listener now reconnects to the database automatically with bounded backoff when the connection drops, instead of needing a manual restart. Would really appreciate a review before the deadline. Per Drips' maintainer guidance, if merging isn't possible in time you can mark the issue resolved anyway so the work still counts. Thanks!