feat: Add health check endpoints for container orchestration

[Edition-X]

Description

Adds health monitoring endpoints to ClearML Serving inference containers for use with Kubernetes, Docker Swarm, and other container orchestration platforms.

New Endpoints

Endpoint	Purpose
GET /health	Basic health check — always 200 when the process is running
GET /readiness	Readiness probe — 503 until ModelRequestProcessor is initialized
GET /liveness	Lightweight liveness probe for orchestration keep-alive checks
GET /health/metrics	JSON service metrics: uptime, request count, loaded models, GPU memory

Changes

clearml_serving/serving/main.py

• Added four health endpoints above
• startup_time tracked at module load for uptime reporting
• instance_id from setup_task() used as the service identifier in /health

clearml_serving/serving/model_request_processor.py

• Added _request_count (guarded by threading.Lock) and _last_prediction_time, incremented in process_request()
• Added get_loaded_endpoints(), get_request_count(), get_last_prediction_time() accessors

Kubernetes Integration

livenessProbe:
  httpGet:
    path: /liveness
    port: 8080
  initialDelaySeconds: 10
  periodSeconds: 30

readinessProbe:
  httpGet:
    path: /readiness
    port: 8080
  initialDelaySeconds: 30
  periodSeconds: 10

Docker Healthcheck

HEALTHCHECK --interval=30s --timeout=10s --retries=3 \
  CMD curl -f http://localhost:8080/health || exit 1

Closes #94

🤖 Generated with Claude Code