Reimplement parallel build feature

[tk0miya]

Is your feature request related to a problem? Please describe.
For now, parallel build does not support Windows platform. Addition to this, it will be disabled soon on macOS and py38+ for security reasons (see #6803 and #6879).
So it would be nice if we'll reimplement the parallel build to support these platform.

Describe the solution you'd like
Reimplement it with "spawn" process model.

Describe alternatives you've considered
N/A

[russellballestrini]

@tk0miya Does parallel builds still work with Linux? What about Linux running py38+?

[tk0miya]

@russellballestrini It is still available. Please let me know if not.

[russellballestrini]

Thank you, I'll test today. I'm working on a project that has 16,000+ documents (huge dataset, I know, so I'll be building my "data pipeline" via docker and linux). I'm leaning hard on the built in javascript index for finding documents quickly without needing a server. Seems to work (with a minor question I raised in the mailing address regarding how to customize the tokenizer)

[Blendify]

Any updates on this? This is a major problem for projects with a large number of files.

[tk0miya]

Nothing improved. It's a chance to contribute! Any PR on this?

[pradyunsg]

it will be disabled soon on macOS and py38+ for security reasons (see #6803 and #6879).

With #9793, MacOS and py38+ can now use the fork model as well.