Feature: Add concurrent library to ParallelComputingInterface

[Bartdoekemeijer]

Feature: add concurrent to ParallelComputingInterface class

This PR adds the ability to use the concurrent library for parallel computing. concurrent is an alternative to the multiprocessing library that is deemed to be more reliable at times, e..g, see the comments in https://stackoverflow.com/questions/20776189/concurrent-futures-vs-multiprocessing-in-python-3.

Related issue

No official issues reported. I have personally experienced some time-out issues with multiprocessing and its interaction with streamlit. Workers may time out due to local errors but that never escalates back up to the main process, and therefore a job may hang forever. Supposedly concurrent is better at handling this.

Impacted areas of the software

Parallel Computing Interface.

Additional supporting information

N/A

Test results, if applicable

The yaw optimization in 12_optimize_yaw_in_parallel.py with interface = "multiprocessing" reports:

===============================================================================
Total time spent for parallel calculation (16 workers): 15.342 s
  Time spent in parallel preprocessing: 0.002 s
  Time spent in parallel loop execution: 15.331 s.
  Time spent in parallel postprocessing: 0.008 s

with interface = "concurrent" I find generally a slight improvement in computation time:

===============================================================================
Total time spent for parallel calculation (16 workers): 14.643 s
  Time spent in parallel preprocessing: 0.002 s
  Time spent in parallel loop execution: 14.632 s.
  Time spent in parallel postprocessing: 0.009 s