Implement Phase 1 task 1.10 (Prometheus metrics + /healthz + /readyz)

Replaces the placeholder Metrics shim with a prom-client implementation
in src/observability/metrics.ts: all 10 Phase 1 metrics from the wiki
spec, plus nodejs_* defaults. Exposes /metrics, /healthz, /readyz over
node:http on METRICS_PORT (9090); /readyz returns 503 when Redis status
is not 'ready' or the TCP listener isn't bound.

The Metrics interface in src/core/types.ts is unchanged — adapter call
sites continue to use the same inc/observe shape. Only main.ts sees the
extended type that adds serializeMetrics().

Side effects:
- Dockerfile re-enables HEALTHCHECK pointing at /readyz, and EXPOSE 9090.
- frame-ingested log downgraded back to debug now that
  teltonika_records_published_total is scrapeable.
- 19 new unit tests covering exposition format, all metric types, and
  every HTTP endpoint path. Total now 98 passing.

Note: deploy/compose.yaml still does not expose 9090 — separate decision
about how Prometheus reaches the service (host port vs. internal scraper
on the same Docker network).

src/main.ts

@@ -1,13 +1,14 @@
 import type { Redis } from 'ioredis';
+import type * as http from 'node:http';
 import type * as net from 'node:net';
 import { loadConfig } from './config/load.js';
 import type { Config } from './config/load.js';
 import { createLogger } from './observability/logger.js';
+import { createMetrics, startMetricsServer } from './observability/metrics.js';
 import { createPublisher, connectRedis } from './core/publish.js';
 import { startServer } from './core/server.js';
 import { createTeltonikaAdapter } from './adapters/teltonika/index.js';
 import { AllowAllAuthority } from './adapters/teltonika/device-authority.js';
-import type { Metrics } from './core/types.js';
 
 // -------------------------------------------------------------------------
 // Startup: validate config (fail fast on bad env), build logger, boot server
@@ -31,12 +32,8 @@ const logger = createLogger({
 
 logger.info('tcp-ingestion starting');
 
-// Placeholder metrics implementation — replaced in task 1.10.
-// Using the Metrics interface from types.ts (no prom-client yet).
-const metrics: Metrics = {
-  inc: (name, labels) => logger.trace({ metric: name, labels }, 'metric inc'),
-  observe: (name, value, labels) => logger.trace({ metric: name, value, labels }, 'metric observe'),
-};
+// Real prom-client metrics implementation (task 1.10).
+const metrics = createMetrics();
 
 // -------------------------------------------------------------------------
 // Wire up the pipeline
@@ -65,10 +62,18 @@ async function main(): Promise<void> {
     metrics,
   });
 
-  // 5. Install graceful shutdown (stub — full hardening in task 1.12)
-  installGracefulShutdown({ server, redis, publisher, logger });
+  // 5. Start metrics HTTP server (task 1.10).
+  // readyzDeps use ioredis's synchronous `.status` field and net.Server's
+  // `.listening` boolean — no I/O, so these closures are always cheap.
+  const metricsServer = startMetricsServer(config.METRICS_PORT, metrics.serializeMetrics, {
+    isRedisReady: () => redis.status === 'ready',
+    isTcpListening: () => server.listening,
+  });
 
-  logger.info({ port: config.TELTONIKA_PORT }, 'tcp-ingestion ready');
+  // 6. Install graceful shutdown (stub — full hardening in task 1.12)
+  installGracefulShutdown({ server, metricsServer, redis, publisher, logger });
+
+  logger.info({ port: config.TELTONIKA_PORT, metricsPort: config.METRICS_PORT }, 'tcp-ingestion ready');
 }
 
 // -------------------------------------------------------------------------
@@ -77,13 +82,14 @@ async function main(): Promise<void> {
 
 type ShutdownDeps = {
   readonly server: net.Server;
+  readonly metricsServer: http.Server;
   readonly redis: Redis;
   readonly publisher: { drain(timeoutMs: number): Promise<void> };
   readonly logger: ReturnType<typeof createLogger>;
 };
 
 function installGracefulShutdown(deps: ShutdownDeps): void {
-  const { server, redis, publisher, logger: log } = deps;
+  const { server, metricsServer, redis, publisher, logger: log } = deps;
 
   let shuttingDown = false;
 
@@ -93,11 +99,15 @@ function installGracefulShutdown(deps: ShutdownDeps): void {
 
     log.info({ signal }, 'shutdown signal received');
 
-    // Stop accepting new connections
+    // Stop accepting new TCP connections
     server.close(() => {
       log.info('TCP server closed');
     });
 
+    // Close the metrics HTTP server before quitting Redis so /readyz reports
+    // not-ready during the drain window (task 1.12 will tighten this further).
+    metricsServer.close();
+
     // Drain publisher queue then disconnect Redis
     publisher
       .drain(10_000)