feat: add HTTP health endpoints to instance manager

[usiegj00]

Summary

Adds HTTP health endpoints to the instance manager, following the CNPG pattern. These endpoints provide health information without spawning processes, enabling future replacement of the current redis-cli based probes.

Fixes #6

Endpoints

Endpoint	Purpose	Returns 200 when	Returns 503 when
/healthz	Liveness	Redis responds to PING	Redis not responding
/readyz	Readiness	Redis ready for traffic	Loading, syncing, master link down
/status	Debugging	Always	Never

Features

• No process spawning: Uses persistent Redis connection
• Cached status: Refreshes every 1 second to minimize Redis load
• Readiness intelligence: Checks loading state, sync progress, master link status
• Configurable: --health-port flag (default 8080)

Response Examples

/healthz

{"status":"ok","redis_pid":123,"uptime_seconds":3600}

/readyz

{"status":"ok","role":"master","connected_clients":5,"loading":false}

/status

{
  "redis": {"pid":123,"role":"master","connected_clients":5,...},
  "replication": {"role":"master","connected_slaves":2,...},
  "instance_manager": {"version":"v1.7.0","uptime_seconds":3600,...}
}

Test Coverage

• Unit tests for all endpoints and helper functions
• E2E tests validating:
  • /healthz returns 200 when healthy
  • /readyz returns 200 when ready
  • /status returns detailed info
  • /healthz returns 503 when Redis is killed
  • Recovery after restart

Test Plan

• Unit tests pass
• E2E tests pass
• /healthz returns 200 for healthy Redis
• /healthz returns 503 when Redis killed
• /readyz returns 200 for ready Redis
• /status returns comprehensive info