THREAD_SUBPROCS=None for xonshrc files

[scopatz]

This is related to the issue at regro/rever#217. I independently verified that even xonsh itself was having trouble starting up if any threaded subproc was used in rc files. Super weird and hard to diagnose. The issue seems to have first arrived in the environment refactor (a8d4a57), but seems to be a side effect of that commit rather than directly caused by it.

[scopatz]

@xonsh/xore - this is ready for review!

[hadim]

I confirm this PR works for me on Ubuntu 20.04 (I am using it only to run rever).

[scopatz]

Pinging this again @xonsh/xore

[laloch]

@scopatz, I hear you. To be completely honest, I don't agree with the change, because it breaks the expected behavior on POSIX. On the other hand it enables job control and removes a lot of confusion about threaded/unthreaded behavior. I would very much like to see a follow-up commit with a documentation change describing thoroughly the effects of the $THREAD_SUBPROCS switch.
I realize, that you are currently pushing towards v1.0, so yes, let's do this, but please, don't forget that this is evil and that we want to find a way to make all our subprocesses reliably threadable.

[scopatz]

Yeah, to be clear I totally agree about wanting to make the sub-processes reliably threadable. Also this doesn't change the default behavior in any way once xonsh is loaded. (ie THREAD_SUBPROCS=True still) This only turns of threadable supbrocs temporarily while xonsh is loading so that you can reliably start xonsh from Bash. Do you still think this needs more docs than I added here?

[scopatz]

Also, thanks for the merge!

[laloch]

Thanks for the PR!

Do you still think this needs more docs than I added here?

Yes, I do. We need to point out, that the RC scripts may behave differently from the regular code and why.

[scopatz]

Ahh Ok I will put it in

[scopatz]

done in #3708