Fix a bug in elpa_multiply_a_b

In elpa_multiply_a_b I wanted to call mpi_comm_rank but accidentally
called mpi_comm_size. This only affects the GPU initialization where
GPUs are assigned to CPU cores based on their MPI ranks.

src/elpa1/elpa_multiply_a_b.F90

@@ -157,7 +157,7 @@
       call mpi_comm_size(int(mpi_comm_rows,kind=MPI_KIND) ,np_rowsMPI ,mpierr)
       call mpi_comm_rank(int(mpi_comm_cols,kind=MPI_KIND) ,my_pcolMPI ,mpierr)
       call mpi_comm_size(int(mpi_comm_cols,kind=MPI_KIND) ,np_colsMPI ,mpierr)
-      call mpi_comm_size(int(mpi_comm_all, kind=MPI_KIND) ,myidMPI ,mpierr)
+      call mpi_comm_rank(int(mpi_comm_all, kind=MPI_KIND) ,myidMPI ,mpierr)
 
       my_prow = int(my_prowMPI,kind=c_int)
       np_rows = int(np_rowsMPI,kind=c_int)