feat: add worker threads, persistent Python sidecar, graceful shutdown, and architectural improvements

- Graceful shutdown: SIGTERM/SIGINT handlers drain HTTP, stop workers, close DB
- Thumbnail caching: disk-cached thumbnails with immutable Cache-Control headers
- Worker thread pool: Piscina offloads Sharp processing off the main event loop
- Persistent Python dispatcher: pre-imports ML libraries, eliminates cold-start latency
- Tool page registry: declarative tool-to-component mapping replaces 750-line switch
- File store cleanup: remove dead derived fields, stable files array reference
- Job persistence: progress written to SQLite jobs table, stale jobs recovered on startup

apps/api/src/lib/worker-pool.ts

@@ -0,0 +1,37 @@
+/**
+ * Worker pool for offloading CPU-bound image processing from the main event loop.
+ *
+ * Uses Piscina (backed by worker_threads) so Sharp operations don't block
+ * HTTP request handling, SSE streams, or health checks.
+ */
+import { availableParallelism } from "node:os";
+import { dirname, resolve } from "node:path";
+import { fileURLToPath } from "node:url";
+import Piscina from "piscina";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+
+// Size the pool: leave 1 thread for the event loop, min 1 worker
+const maxThreads = Math.max(1, Math.min(availableParallelism() - 1, 4));
+
+let pool: Piscina | null = null;
+
+export function getWorkerPool(): Piscina {
+  if (!pool) {
+    pool = new Piscina({
+      filename: resolve(__dirname, "image-worker.ts"),
+      // Inherit tsx loader flags from the main process so .ts files work in workers
+      execArgv: [...process.execArgv],
+      maxThreads,
+      idleTimeout: 30000,
+    });
+  }
+  return pool;
+}
+
+export async function shutdownWorkerPool(): Promise<void> {
+  if (pool) {
+    await pool.destroy();
+    pool = null;
+  }
+}