MARS-145 follow-up: daytime-peak 53300 re-scan of the txn-pooler runtime pool

MARS-145 (c987038) cut the runtime pool over to the Supavisor txn-mode pooler (:6543) + lowered pg.Pool max 8->6. Post-push PTD landed clean but at ~04:05 ART = near-zero traffic: the CONNECT path is verified (drain-email-outbox 200 roundtrip through the pooler) but the pool CEILING under concurrent daytime load is NOT exercised. Absence of 53300/'too many'/'remaining connection slots' overnight is not proof it can't happen at peak. TASK (db-mars or audit): during tomorrow's daytime peak window, sweep appEvents + Vercel runtime logs for 53300 / connection-slot-exhaustion / pooler connect failures attributable to the running deployment. If clean at peak -> close, cutover fully validated. If 53300 appears -> the client max:6 vs Supavisor default_pool_size interaction needs revisiting (raise max, or tune the server-side pool). Owner = audit/db monitoring, not a code change unless load proves the sizing wrong.