fix 32-bit/64-bit overflow issue in expression

  // post all receives, starting after self copies

  bigint offset = num_self*nbytes;

  bigint offset = num_self*(bigint)nbytes;

  for (int irecv = 0; irecv < nrecv_proc; irecv++) {

    MPI_Irecv(&recvbuf[offset],num_recv[irecv]*nbytes,MPI_CHAR,

              proc_recv[irecv],0,world,&request[irecv]);

  // wait on all incoming messages

  if (nrecv_proc) MPI_Waitall(nrecv_proc,request,status);

  // approximate memory tally

  // DEBUG lines

  //bigint irregular_bytes = 2*nprocs*sizeof(int);

  //irregular_bytes += maxindex*sizeof(int);

  //irregular_bytes += maxbuf;

}

/* ----------------------------------------------------------------------