We should add tests to ensure that the JDBC_PING2 protocol behaves as expected with all of our supported DBs. More specifically, we should ensure that cluster membership works as expected pre/post DB disconnection and when the cluster is scaled up/down due to abnormal server termination.

This is necessary to ensure that Infinispan works well with JDBC_PING2, as Keylcoak intends to make this their default ping protocol for non Kubernetes environments.