Allow specifying stdout to Popen

[s0undt3ch]

No description provided.

[s0undt3ch]

The reason behind this change is that on a particular session, we need access to the output, even if the command failed, for CI reporting reasons.

[theacodes]

Can you show me an example of when you'd use this and why the existing mechanisms aren't giving you the output you expect?

[mostealth]

When running in CI you might want to save pylint output to file (to be later parsed). pylint does not support outputting to file hence the only solution I found was to monkey patch subprocess.Popen so that I inject my own stdout/stderr file before calling session.run

[s0undt3ch]

What @mostealth said, that's our exact same scenario and why I created this PR.

This solution in use on our repo can be seen here

[theacodes]

That seems like a great reason. I had a comment about covering some edge cases, but I'm happy to merge this once we get that sorted. :)

[s0undt3ch]

Rebased and added two new tests based on the comments provided.

[theacodes]

Okay, I made silent and stdout mutually exclusive since silent doesn't have an effect if stdout is passed.

Does this still look good to you, @s0undt3ch? Also, it should be technically possible to override stderr now. Do you mind adding a test for that?

[s0undt3ch]

Added tests for custom stderr

[theacodes]

Thank you, @s0undt3ch!

[s0undt3ch]

Anytime!