Collect health status of watched services asynchronously and regularly

[sganslandt]

... and simple respond with the lates fetched status when you request them from /healths

• [x] move health checking out of HealthsResource and collect asynch
• [x] check all services individually and not in a single loop
• [x] only publish events if the health actually change
• [ ] include status about which direction it has changed in