Reduce backend per-upload CPU to relieve upload timeouts

The Node main thread was saturated at about 39 uploads/s, overflowing the
listen queue so uploads timed out connecting from nginx.

- Enable keepAlive on the S3 agent; every upload did a full TLS handshake
- Remove OpenTelemetry tracing
- Cache the system key lookup for 60s
- Use lean queries when listing calls
- Serialize each call once in notify_clients, not per recipient
- Add an nginx upstream keepalive pool, and hold Node's keepAliveTimeout
  above it so nginx never reuses a connection Node has closed

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

backend/agents/otel-tracing.js

@@ -1,69 +0,0 @@
-'use strict';
-
-const opentelemetry = require('@opentelemetry/sdk-node');
-const { MongoDBInstrumentation } = require('@opentelemetry/instrumentation-mongodb');
-const { TraceIdRatioBasedSampler } = require('@opentelemetry/sdk-trace-node');
-const { OTLPTraceExporter } =  require('@opentelemetry/exporter-trace-otlp-http');
-const { getNodeAutoInstrumentations } = require('@opentelemetry/auto-instrumentations-node');
-const { diag, DiagConsoleLogger, DiagLogLevel } = require('@opentelemetry/api');
-const {
-  Detector,
-  DetectorSync,
-  IResource,
-  ResourceDetectionConfig,
-  envDetectorSync,
-  hostDetectorSync,
-  processDetectorSync,
-} = require("@opentelemetry/resources")
-// const {
-//     dockerCGroupV1Detector,
-//   } = require('@opentelemetry/resource-detector-docker');
-
-// For troubleshooting, set the log level to DiagLogLevel.DEBUG
-diag.setLogger(new DiagConsoleLogger(), DiagLogLevel.WARN);
-
-console.log("Starting OpenTelemetry Node SDK");
-console.log("OTEL_EXPORTER_OTLP_ENDPOINT: " + process.env.OTEL_EXPORTER_OTLP_ENDPOINT);
-console.log("OTEL_EXPORTER_OTLP_HEADERS: " + process.env.OTEL_EXPORTER_OTLP_HEADERS);
-console.log("OTEL_SERVICE_NAME: " + process.env.OTEL_SERVICE_NAME);
-
-
-// from https://github.com/open-telemetry/opentelemetry-js/issues/4638
-// this is a workaround to wait for async attributes to be resolved
-function awaitAttributes(detector){
-  return {
-    async detect(config) {
-      const resource = detector.detect(config)
-      await resource.waitForAsyncAttributes()
-
-      return resource
-    },
-  }
-}
-
-const samplePercentage = 0.05;
-
-const sdk = new opentelemetry.NodeSDK({
-    traceExporter: new OTLPTraceExporter(),
-    sampler: new TraceIdRatioBasedSampler(samplePercentage),
-    instrumentations: [
-        new MongoDBInstrumentation({
-            // see under for available configuration
-          }),
-        getNodeAutoInstrumentations({
-            // we recommend disabling fs autoinstrumentation since it can be noisy
-            // and expensive during startup
-            '@opentelemetry/instrumentation-fs': {
-                enabled: false,
-            },
-        }),
-    ],
-    resourceDetectors: [
-      awaitAttributes(envDetectorSync),
-      awaitAttributes(processDetectorSync),
-      awaitAttributes(hostDetectorSync),
-    ],
-  //resourceDetectors: [dockerCGroupV1Detector],
-});
-
-sdk.start();