Add DB retry/backoff and improve logging

[scoop1120]

Improve resilience and observability around DB access and GraphQL requests.

• Add environment constants EXO_DB_RETRY_MAX, EXO_DB_RETRY_BASE_MS, EXO_DB_RETRY_MAX_MS.
• Implement retry/backoff logic for read-only (SELECT) SQL operations with configurable max retries, base and max backoff, jitter, and detailed error logging for Postgres errors.
• Add RetryConfig and helper functions to detect transient SQL errors and compute backoff.
• Log and map transaction start failures to DatabaseError with tracing error context.
• Warn and drop closed DB client in TransactionState before reacquiring.
• Emit detailed error info when failing to acquire a pooled DB client (pool status + error) instead of silently propagating.
• In GraphQL router: extract request ID from common headers, add it to Sentry scope, create a tracing span per request and log receipt.

These changes aim to handle transient DB connectivity issues more robustly and give better diagnostics for DB and request lifecycle failures.

[github-actions]

Version bump detected: 0.30.1 → 0.30.2