ClusterPartition and related services must share the same multiplexed JChannel

ClusterPartition/HAPartition expose DistributedReplicantManager and DistributedState as part of their API. These 2 services use the "ClusterTreeCache" JBC instance to manage state. To allow consistent use of HAPartition and DRM/DS, they must be talking to the same underlying JChannel. To ensure this:

1) ClusterTreeCache will be injected into ClusterPartition.
2) ClusterPartiton will check that ClusterTreeCache is using the multiplexer. It will obtain it's multiplexer reference and stack name from ClusterTreeCache.
3) ClusterPartition will no longer support creation of a JChannel from a jgroups protocol stack config element.