Knora should recover from temporary failure to access the triplestore

[subotic]

When the triplestore is not available after the Knora service successfully started, following should happen:

• change state to unhealthy
• answer every API request with a 503 Service Unavailable
• start periodic checking for the triplestore (much like at startup)
• after triplestore is available again, maybe initiate some deeper checking:
  • reload ontologies?
  • trigger GraphDB consistency checker?
• change state to healthy
• start handling API requests again.

[subotic]

maybe useful: https://doc.akka.io/docs/akka/current/common/circuitbreaker.html