Use multiprocessing in the file discovery

[ericwb]

Is your feature request related to a problem? Please describe.
To boost performance, the file discovery could use multiprocessing.

Describe the solution you'd like
Create a process pool to build list of files.

discover_files() currently takes around 0.51 seconds to discovery all files in cpython.

Describe alternatives you've considered
Single process

Additional context
n/a

Love this idea? Give it a 👍. We prioritize fulfilling features with the most 👍.