Run local operations from a DistributedExecutor in parallel

Looking into

org.infinispan.distexec.DefaultExecutorService.invokeLocally(DistributedRunnableFuture<T>)

It seems that the operations are performed in the same thread as the invoker; if we happen to send multiple operations to the local node, these are performed in sequence, and in sequence with the remote operations. We should use an Executor for the local tasks.

Edit by Manik:
After discussing with Sanne and Vladimir, I am reopening this issue as it isn't really a duplicate of ISPN-1274. This is more about being able to take a task and run the task, locally, using multiple threads (not just on a separate thread). Essentially a lot closer to Fork/Join, but with the ability to run this on multiple nodes. The degenerate case would possibly look a lot like the JDK's F/J, on a single node.