[Feature Request] Provide worker status

[cretz]

Describe the solution you'd like

Users need to get worker status to at least check worker health and programmatically react to information we often only make available via logs/metrics. This should be a common design applied to all SDKs. What should be made available (all subject to discussion):

• Current activity slots in use
• Current workflow slots in use (debatable since this comes and goes so quick)
• Current workflow cache size (in SDKs that aren't using global caches)
• Current status per worker type - e.g. "polling", "full", or "failure backoff"
  • Make sure failure backoff includes the error
  • Consider way to notify user programmatically upon status change
• Server reachable (i.e. that client doesn't have any server errors)

Per-SDK Tickets

• Go - TODO
• Java - TODO
• Core - TODO
• TypeScript - TODO
• Python - TODO
• .NET - TODO
• Ruby - TODO
• PHP - TODO
• Temporal CLI - TODO

[dannymcerlean1]

+1. This would be really helpful for monitoring.

[yuandrew]

Created a separate, related issue for worker heartbeating, #696, and reverting the changes I recently made for this issue. I didn't realize this issue was specifically regarding getting the status from the worker directly, whereas worker heartbeating queries server on the workers its aware of.