Implement Retries on Transient Roomlog DB Connection Errors

[DieterReinert]

Problem:
Previously, any unexpected DB connection termination resulted in lost logs, since we only retried on 42P01 (table not found). Other transient failures (e.g., “Connection terminated unexpectedly”) were crashlogged without a retry, causing data loss.

Solution:
Introduce a bounded retry mechanism (3 attempts) specifically for transient “Connection terminated unexpectedly” errors. After a brief wait, the query is re-run. If retries are exhausted, or if ignoreFailure is true, the function skips retries and relies on crashlogs as before.

Benefits:

• Preserves Logs: Prevents loss of log entries due to DB connection drops by retrying the insert.
• Consistent Recovery Logic: Extends existing table recreation and retry logic to handle additional transient errors.
• Respecting ignoreFailure: Honors the ignoreFailure flag to bypass retries when log insertion isn't critical.
• Minimal Performance Impact: Implements a bounded retry with short delays, avoiding excessive resource usage.
• Enhanced Debugging: Continues to crashlog genuine, non-transient errors, ensuring visibility into persistent issues.