Fix setting MPI processes and ranks (#1808)

0.12.0 accidentally removed MPI support if application does not use qobj. This commit reverts the change.

releasenotes/notes/fix_mpi_procs-68b76c11fe7a6b8e.yaml

@@ -0,0 +1,5 @@
+---
+fixes:
+  - |
+    MPI parallelization was not enabled since we have not used qobj.
+    This fix sets the number of processes and MPI rank correctly.

src/controllers/aer_controller.hpp

@@ -852,11 +852,6 @@ Transpile::CacheBlocking Controller::transpile_cache_blocking(
 //-------------------------------------------------------------------------
 template <typename inputdata_t>
 Result Controller::execute(const inputdata_t &input_qobj) {
-#ifdef AER_MPI
-  MPI_Comm_size(MPI_COMM_WORLD, &num_processes_);
-  MPI_Comm_rank(MPI_COMM_WORLD, &myrank_);
-#endif
-
   // Load QOBJ in a try block so we can catch parsing errors and still return
   // a valid JSON output containing the error message.
   try {
@@ -908,6 +903,10 @@ Result Controller::execute(std::vector<Circuit> &circuits,
   // Start QOBJ timer
   auto timer_start = myclock_t::now();
 
+#ifdef AER_MPI
+  MPI_Comm_size(MPI_COMM_WORLD, &num_processes_);
+  MPI_Comm_rank(MPI_COMM_WORLD, &myrank_);
+#endif
   // Determine simulation method for each circuit
   // and enable required noise sampling methods
   auto methods = simulation_methods(circuits, noise_model);