Add puma metrics to prometheus

Goals

As part of the experimental puma work, it's essential that we can monitor the new server

• Ensure that important puma metrics are exported via the prometheus endpoint
• These should include:
  • Memory consumption (rss, vmm?) for worker processes and master process
  • Stats provided by Puma workers, phase, booted_workers, old_workers, worker_status
  • Thread count, open fd count, etc