avoid multiprocessing when simulating only one Task

[RamiALBASHA]

RuFaS currently instantiates a multiprocessing.Pool for running the simulation Tasks, regardless of the number of the latter. In the case of running only one Task, using multiprocessing.Pool is not necessary and can be replaced by a sequential calculation. This replacement is beneficial for users who deploy the code to AWS Lambda services (like our team) where multiprocessing with Pool is not supported.

This feature can be performed with minimal modifications to the code, by:

1. assigning self.pool to None if the number of workers is 1 (here):

self.pool = multiprocessing.Pool(workers, maxtasksperchild=1) if workers > 1 else None

2. switching between parallel or sequential processing depending on the value of pool (here)

if self.pool is not None:
    results = self.pool.imap(task_with_args, single_run_args)
else:
    results = (task_with_args(v) for v in single_run_args)

We have implemented these modifications to our fork and wanted to share the info for the benefit of other users.

[PooyaHekmati]

Hi Rami,
Thanks for opening this discussion. I think this can be a good addition to the codebase. We would appreciate it if you could open a PR please. Before that, however, I think it is a good idea to double check with the other teams that are using RuFaS to make sure we won't break things for them. @JoeWaddell would you please ask folks that you are working with about this?

[JoeWaddell]

I was able to confirm this won't affect the workflow the other team is currently using, so thanks to you both!

[RamiALBASHA]

Ok, I'll open a PR for this modification.

[RamiALBASHA]

Here's the PR:
#2685