fix deletion of deployments in cluster

when a deployment gets deleted, the request arrives in one node of the cluster. that node can remove the deployment and its process definition from the repository cache immediately.

but how will the other nodes in the cluster know ?

some check with db access has to be added every time an object (process definition) is fetched from the repository cache. alternatively this can also be enforced at those locations where suspended must be checked: in the commands for starting a new execution, signalling and so on.