load balancing and health checks for remote chacra instances

[alfredodeza]

Configured remote chacra instances to be load balanced should have a health check implementation.

Ideally this could be accomplished by an asynchronous worker and a broker, but this would involve adding extra complexity/maintenance.

Since chacra is going to implement a callback system, that system could support a health check ping (see chacra/issues/110 ).

The check would need at least a configurable timeout value between checks. This would enable a failed chacra instance that no longer pings to be removed from the configured group in shaman.

The domain logic of what constitutes a healthy chacra instance should be handled by chacra itself.

[alfredodeza]

Load balancing is now complete.

Health checks are not enabled (nginx will proxy regardless)