Add warning log to validator client if primary node is unhealthy

[nflaig]

Problem description

Right now, the validator client is pretty silent about the primary node being unhealthy. We recently added metrics and dashboard panels to show the status of connected nodes (#6415) but we only log requests to fallback nodes to debug which many users do not look at.

Solution description

Add warning log to validator client if primary node is unhealthy as suggested by seamonkey.

We could use the same approach to flag the primary node as unhealthy if score is at 0. This would prevent false positives if just a single request fails which might happen due to other reasons.

Additional context

Related to #6415