A single Auth0 failure should be an info, not a warning

[johnml1135]

We got this warning:
Health check Auth0 with status Unhealthy completed after 13.2294ms with message 'Discover endpoint is not responding with 200 OK, the current status is InternalServerError and the content \r\n 500

If there are a few failures (such as they are doing a server switchover), we should not ping on it. We should implement a certain "fault tolerance" similar to how we do for ClearML.