Fix sync server startup against a sleeping database

[lobbystack]

The server crashed into a restart loop when its Postgres schema probe ran while the managed database was still waking after scale-to-zero. Railway logs showed repeated connect ETIMEDOUT/connect ECONNREFUSED and uncaught-error restarts.

• Add bounded startup retry with exponential backoff around store.ready() (60 attempts, 1s doubling to a 5s cap) so the process waits for the database instead of crashing.
• Set connect_timeout: 10 on the postgres client so each failed TCP attempt fails fast.
• Patch Better Auth's Postgres type allowlist to recognise the internal int8/int2/float4/float8 type names, silencing a false rateLimit.lastRequest type warning. The column is correctly bigint (epoch milliseconds); no schema change.
• Add provenance for the patch and regression tests for both changes.

Tests: startup retry unit tests, dependency-patch guard, existing server suite.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.