Coprocessor health check as part of overall router health

[lleadbet]

Is your feature request related to a problem? Please describe.

The router currently fails closed if the coprocessor is unavailable/erroring/etc, which leads to situations where a router may be marked as healthy by it's own health check, but not by clients.

Describe the solution you'd like

Ideally coprocessors, when enabled, are used when calculating the overall health of the router. As an example shape of a possible config:

coprocessor:
  url: http://localhost:3000/
  healthcheck:
    enabled: true
    endpoint: "/health"

Which would use a user-defined health endpoint for the coprocessor.

Describe alternatives you've considered

Moving all health checks into the coprocessor instead, when running in a sidecar model (i.e. router and coprocessor exist in the same Kubernetes pod).