fix(forge_repo): retry database pool creation and connection access

[amitksingh1490]

Summary

Retry SQLite database pool creation and connection checkout when .forge.db is temporarily unavailable, such as when another process is holding the database file.

Context

Forge currently fails hard when the SQLite pool cannot be created immediately. This change makes database access more resilient by retrying transient pool creation and pool checkout failures instead of failing on the first attempt.

Changes

• add backon to forge_repo for blocking retry support
• add configurable max_retries to PoolConfig
• reduce the default pool connection timeout from 30s to 5s so retries happen sooner
• retry database pool creation with exponential backoff
• retry connection checkout from the pool with the same backoff strategy
• store retry configuration on DatabasePool so runtime checkouts use the configured retry count
• keep the shared retry behavior in one helper to avoid duplication

Testing

• cargo check -p forge_repo
• cargo test -p forge_repo -- --test-threads=1