Run server and client(s) in different containers

In the module lra/test, the separation between the server (i.e. lra-coordinator) and clients (lra-participant(s) [e.g. LRAParticipant, LRAParticipantWithoutStatusURI, LRAParticipantWithStatusURI, etc.]) might benefit from a separation of roles. In fact, using two different containers to run server and client separately would be closer to a production scenario.

To implement this scenario, the deployment mechanism in lra/test should be extended and an Arquillian cluster should be considered.