Feature: Failback after failover

Thank you for this  tool!

A useful feature would be polling endpoints that failed, to see whether they can resume duty, and fail back. This would implement a hierarchy of endpoints, where endpoints specified earlier are always preferred over endpoints specified later.

Currently, failback requires a restart of the proxy.
