Infinispan cache is definitely desynchronized after network trouble

Issue

• Network failure causes temporary cluster split and re-join
• During the cluster split at least some nodes can still process requests, so that local cache content changes
• After the cluster member re-join the defined Infinispan replicated-caches don't perform a merge
• Results of this behavior: The content of every replicated-cache becomes un-reliable when
  • a cluster splits due to network trouble
  • stays un-reliable even after a re-join, as no merge is triggered.