feat: restore Langfuse observability integration (#103)

- Add lib/langfuse.ts with client, trace input/output, telemetry config
- Add instrumentation.ts for OpenTelemetry setup with Langfuse span processor
- Add /api/log-save endpoint for logging diagram saves
- Add /api/log-feedback endpoint for thumbs up/down feedback
- Update chat route with sessionId tracking and telemetry
- Add feedback buttons (thumbs up/down) to chat messages
- Add sessionId tracking throughout the app
- Update env.example with Langfuse configuration
- Add @langfuse/client, @langfuse/otel, @langfuse/tracing, @opentelemetry/sdk-trace-node

instrumentation.ts

@@ -0,0 +1,35 @@
+import { LangfuseSpanProcessor } from '@langfuse/otel';
+import { NodeTracerProvider } from '@opentelemetry/sdk-trace-node';
+
+export function register() {
+  // Skip telemetry if Langfuse env vars are not configured
+  if (!process.env.LANGFUSE_PUBLIC_KEY || !process.env.LANGFUSE_SECRET_KEY) {
+    console.warn('[Langfuse] Environment variables not configured - telemetry disabled');
+    return;
+  }
+
+  const langfuseSpanProcessor = new LangfuseSpanProcessor({
+    publicKey: process.env.LANGFUSE_PUBLIC_KEY,
+    secretKey: process.env.LANGFUSE_SECRET_KEY,
+    baseUrl: process.env.LANGFUSE_BASEURL,
+    // Filter out Next.js HTTP request spans so AI SDK spans become root traces
+    shouldExportSpan: ({ otelSpan }) => {
+      const spanName = otelSpan.name;
+      // Skip Next.js HTTP infrastructure spans
+      if (spanName.startsWith('POST /') ||
+          spanName.startsWith('GET /') ||
+          spanName.includes('BaseServer') ||
+          spanName.includes('handleRequest')) {
+        return false;
+      }
+      return true;
+    },
+  });
+
+  const tracerProvider = new NodeTracerProvider({
+    spanProcessors: [langfuseSpanProcessor],
+  });
+
+  // Register globally so AI SDK's telemetry also uses this processor
+  tracerProvider.register();
+}