Provide a standardized way to run tests from cmdline

[giampaolo]

This idea originated from the discussion at #987.
Right now the recommended way to run tests is to get a copy of the source distribution (namely via GIT) and run make test which will:

1. install psutil
2. install testing deps (unittest2, mock, pywin32 etc.)
3. run tests

It would be nice, though, to provide a way to quickly test the psutil installation via cmdline and have it documented. Right now, the undocumented way is:

python -c "from psutil.tests.runner import main; main()"

...but it's a bit twisted. CPython accomplishes this with (https://docs.python.org/devguide/runtests.html):

python -m test

It would be nice if psutil did the same:

python -m psutil.test

This should be implemented and documented in the official doc as a quick way to make sure whether psutil works on the designated user platform (aka running tests) without cloning the GIT repo or downloading the tarball.

Currently the GIT cloned code / tarball allows you to make test and make setup-dev-env in order to install the testing deps (unittest2, mock, pywin32, etc.), which is another useful use case, so we should also be able to:

python -m psutil.test --install-deps

If python -m psutil.test fails for a missing dep, the error message should tell the user to use --install-deps first.

[giampaolo]

PR: #1019