Add status.json endpoint to health check endpoints

[moskyb]

Description

This PR adds a new endpoint to the healthcheck server, /status.json. the idea is that this endpoint will be a very cut-down version of the information available at /status, but in a format (JSON) that's easily machine-readable.

A sample of this output is:

{
  "health": "ok",
  "aggregate_status": "busy", # One of the workers below is busy, so the "aggregate status" of the agent is busy
  "worker_statuses": [
    {
      "status": "idle",
      "id": "018f2d68-1d0c-42cd-be6c-c1a89bda6453",
      "spawn_index": 1
    },
    {
      "status": "idle",
      "id": "018f2d68-1e33-4f09-ae1e-8b962f4e4e77",
      "spawn_index": 2
    },
    {
      "status": "busy",
      "current_job_id": "018f2d68-5e54-4255-a99e-829aa4691dc1", # only busy agents have a current job ID
      "id": "018f2d68-1f57-4374-9dbe-ec033f915c75",
      "spawn_index": 3
    },
    {
      "status": "idle",
      "id": "018f2d68-2080-4eaf-a33d-e1c3d0521206",
      "spawn_index": 4
    }
  ]
}

This PR also refactors the healthcheck server to be fully owned by the AgentPool, and replaces the use of the default mux (used when http.HandleFunc is called) with a dedicated mux for just the healthcheck server.

Testing

• Tests have run locally (with go test ./...). Buildkite employees may check this if the pipeline has run automatically.
• Code is formatted (with go fmt ./...)

[DrJosh9000]

Looks pretty good

[DrJosh9000]

🥒