[Cluster] Improve message passing part.

This rewrites in part the message passing part, using in place gather, and collecting, rather than merging, the data of all threads.

neutral with a single thread per rank:
Score of new-2mpi-1t vs old-2mpi-1t: 789 - 787 - 2615  [0.500] 4191
Elo difference: 0.17 +/- 6.44

likely progress with multiple threads per rank:
Score of new-2mpi-36t vs old-2mpi-36t: 76 - 53 - 471  [0.519] 600
Elo difference: 13.32 +/- 12.85

src/thread.cpp

@@ -139,6 +139,9 @@ void ThreadPool::set(size_t requested) {
 
       // Reallocate the hash with the new threadpool size
       TT.resize(Options["Hash"]);
+
+      // Adjust cluster buffers
+      Cluster::ttRecvBuff_resize(requested);
   }
 }