--user fixes part5: reset_env option for adding patches to sitecustomize

[qwcode]

part 5 of the pull #511 breakup
https://gist.github.com/2822510

part 5 is new in the plan as of today.

in writing tests for the next part, I was needing support for adding a monkey patch snippet to sitecustomize.py

this change adds an option to reset_env to allow any test to add a python code snippet to sitecustomize.py.

if you recall from part1, PyPIProxy.setup (which applies a monkey patch to urlopen) was moved to sitecustomize.py. the method that places that code was refactored to make use of this new general method.

the tests I wrote for this should make clear what this is about.

[travisbot]

This pull request fails (merged 6934084 into ae867db).

[pnasrat]

Can you check the failures on 3.2

Test the test support awesomeness though.

[qwcode]

my local tox py32 run doesn't fail on the test that Travis fails on.
looking at the assertion though to try to understand what would lead to that state...

[travisbot]

This pull request passes (merged 1d1363b into ae867db).

[qwcode]

hmm, I changed nothing relevant (expect to let errors bubble up), and it passed this time?

[qwcode]

I confirmed that there is not redundant sitecustomize growth in the Fast environment.
it's created after the backup is made that rebuilds the environment.
but I still made some changes to make it less likely that a refactor might make it start happening, and added a test to confirm that it's not happening

[travisbot]

This pull request passes (merged faee69b into ae867db).

[qwcode]

so closed, but not merged?
how do I move forward with the plan?

[pnasrat]

Umm I hit merge, not sure what happened

[travisbot]

This pull request passes (merged faee69b into ae867db).

[qwcode]

yes, it worked!!