fix: database connection resilience — retry on transient errors, TCP keepalive, health check timeout

- Enable TCP keepalive on pg.Pool to detect dead connections
- Add connectionTimeoutMillis (5s) to prevent hanging on stale connections
- Add queryWithRetry() with exponential backoff for transient DB errors
- Add connectWithRetry() for transaction-based operations
- Detect PgBouncer "no available server" and other transient errors
- Health check has 3s timeout and returns 503 on DB failure
- All DB operations in keys, verification, usage use retry logic

Fixes BUG-075: PgBouncer failover causes permanent pod failures

src/middleware/usage.ts

@@ -1,6 +1,7 @@
 import { isProKey } from "../services/keys.js";
 import logger from "../services/logger.js";
 import pool from "../services/db.js";
+import { queryWithRetry, connectWithRetry } from "../services/db.js";
 
 const FREE_TIER_LIMIT = 100;
 const PRO_TIER_LIMIT = 5000;
@@ -22,7 +23,7 @@ function getMonthKey(): string {
 
 export async function loadUsageData(): Promise<void> {
   try {
-    const result = await pool.query("SELECT key, count, month_key FROM usage");
+    const result = await queryWithRetry("SELECT key, count, month_key FROM usage");
     usage = new Map();
     for (const row of result.rows) {
       usage.set(row.key, { count: row.count, monthKey: row.month_key });
@@ -40,7 +41,7 @@ async function flushDirtyEntries(): Promise<void> {
 
   const keysToFlush = [...dirtyKeys];
   
-  const client = await pool.connect();
+  const client = await connectWithRetry();
   try {
     await client.query("BEGIN");
     for (const key of keysToFlush) {