Use multiprocessing to speed up processing of input files

[hugovk]

Use multiprocessing to speed up processing of input files, using the number of processes which matches the CPU count.

On my old dual-core Mac, the processing step speeds up:

• 601 files (all of 2021): 2m7s -> 1m20s
• 2,629 files (my full Strava archive): 10m29s -> 5m9s

Will be a larger gain for machines with more CPUs.

multiprocessing needs a single worker function, this is the new process_file which decides which of process_gpx and process_fit to call. They also need to be top-level functions to work with multiprocessing.

I also formatted the file using the popular Black autoformatter:

python -m pip install -U black
black . 

---

I also fixed a bug when using an input path like ~/dir, it needs to expand ~ before checking it's a dir (and appending *).

[marcusvolz]

Very cool.