Add DB pooling, throttling and monitor backoff

Add connection pool params to DATABASE_URL and configure Prisma to use the pooled URL (connection_limit=20, pool_timeout=30) to reduce connection exhaustion. Introduce safeguards and throttling across processors: limit in-flight progress DB updates in direct-download, add short delays when processing RSS, retry-failed-imports, and retry-missing-torrents, and stagger scheduler triggers to avoid bursts. Implement adaptive monitor-download polling with stallCount/lastProgress and exponential backoff, and thread these fields through JobQueueService (including reduced worker concurrency for several queues). Batch audiobook enrichment queries to small parallel batches to limit DB load. Update tests to reflect new monitor payload parameters. Overall intent: reduce DB connection pool pressure and smooth load spikes during startup and heavy processing.

src/lib/processors/retry-missing-torrents.processor.ts

@@ -44,6 +44,7 @@ export async function processRetryMissingTorrents(payload: RetryMissingTorrentsP
     }
 
     // Trigger appropriate search job for each request based on type
+    // Throttle: 100ms delay between jobs to avoid connection pool burst
     const jobQueue = getJobQueueService();
     let triggered = 0;
 
@@ -73,6 +74,9 @@ export async function processRetryMissingTorrents(payload: RetryMissingTorrentsP
       } catch (error) {
         logger.error(`Failed to trigger search for request ${request.id}: ${error instanceof Error ? error.message : 'Unknown error'}`);
       }
+
+      // Spread DB operations over time to avoid connection pool exhaustion
+      await new Promise(resolve => setTimeout(resolve, 100));
     }
 
     logger.info(`Triggered ${triggered}/${requests.length} search jobs`);