Compute plot in parallel

This MR allows to compute plot in parallel, should only be merge once the memory usage of plot generation have be reduced

jass/tasks.py

@@ -74,31 +74,38 @@ def create_project_csv_file(project_id):
     return project.create_csv_file()
 
 
+@celery.task
+def dummy_task():
+    print("This task to nothing, but help the chain to have a valid status")
+
+
 def run_project_analysis_if_needed(project):
     if project.has_computation_going_on():
         return
     tasks = []
+    post_worktable_tasks = []
     if not os.path.exists(project.get_worktable_path()):
         tasks.append(create_project_worktable.si(project.id))
 
     if project.get_global_manhattan_plot_path() and not os.path.exists(project.get_global_manhattan_plot_path()):
-        tasks.append(create_project_global_plot.si(project.id))
+        post_worktable_tasks.append(create_project_global_plot.si(project.id))
 
     if project.get_quadrant_plot_path() and not os.path.exists(project.get_quadrant_plot_path()):
-        tasks.append(create_project_quadrant_plot.si(project.id))
+        post_worktable_tasks.append(create_project_quadrant_plot.si(project.id))
 
     if project.get_qq_plot_path() and not os.path.exists(project.get_qq_plot_path()):
-        tasks.append(create_project_qq_plot.si(project.id))
+        post_worktable_tasks.append(create_project_qq_plot.si(project.id))
 
     if project.get_csv_path() and not os.path.exists(project.get_csv_path()):
-        tasks.append(create_project_csv_file.si(project.id))
+        post_worktable_tasks.append(create_project_csv_file.si(project.id))
 
-    if len(tasks) == 0:
+    if len(tasks) + len(post_worktable_tasks) == 0:
         return
     main_wf = chain(
         *tasks,
+        group(post_worktable_tasks),
         # it is mandatory to add a task that do nothing if group is used in order to have a success/failure status
-        # dummy_task.si(),
+        dummy_task.si(),
     )
     task = main_wf.delay()
     project.celery_id = task.id