Add health checks

(To be rebased after cleanup-v2 is merged) Add backend onion service health checking. If an onion service becomes unreachable it is removed from the fronted onion service. It is added back automatically as needed. The events are logged as info or error level. It supports checking TCP, HTTP and HTTPS. By default the checking is disabled.

Timeouts and check intervals are hardcoded and could be made configurable in a later MR.