better way to handle mpi

aggregator.py

@@ -4,16 +4,13 @@ import numpy as np
 from functools import partial
 from baseprocessor import Rule
 import snapshotprocessor
-
+from mypool import MyPool
 
 try:
     from mpi4py.futures import MPIPoolExecutor
-
+    mpi_loaded = True
-    mpi = True
 except ModuleNotFoundError:
-    from mypool import MyPool
+    mpi_loaded = False
-
-    mpi = False
 
 
 def _map_aux(fun, path, path_out, params, run_num, **kwargs):
@@ -57,7 +54,7 @@ class Aggregator:
             map_fn = partial(
                 _map_aux, func, self.path, self.path_out, self.params, **kwargs
             )
-            if mpi:
+            if mpi_loaded and self.params.process.mpi:
                 executor = MPIPoolExecutor(max_workers=num_process)
                 try:
                     result = list(executor.map(map_fn, run_num, unordered=True))

params.yml

@@ -96,6 +96,8 @@ out: # Parameters for post processing
 process: # General setting of the post-processor module
     verbose : True                     # Give more infos on what is going on
     num_process : 1                    # Number of forks
+    mpi         : False                # If available, use mpi4py
+    allow_error : False                # Catch all errors in rules involving the disk
     save_cells : True                  # Save cells structure on disk
     save_parts : True                  # Save particles on disk
     unload_cells : True                # Save memory usage
@@ -105,6 +107,7 @@ rules: # Specific rules parameters
 
 
 astrophysix: # Parameters for astrophysix and galactica
+    generate          : False              # Whether to generate astrophysix objects
     simu_fmt          : "{tag}_{run}"      # Format of the name of simulation
     descr_fmt         : "{tag}_{run}"      # Format of the default description
     # The following keys are accepted