This is an emerging best practice and there is little reason to not
follow it

Add a version info tuple for plugins to use

This relies on two things:

1. Properly configuring flake8-import-order to use that style
2. Properly configuring flake8-import-order to know that flake8 is our
   application name.

ConfigFileFinder should absolutely handle broken/invalid config files
by refusing to try to parse them. Here we catch the ParsingError,
log the exception, and then return normally. The RawConfigParser
instance is perfectly valid still and will behave as if nothing had
been read and we just need to indicate that we didn't find any files
worthy of reading.

Related to: https://github.com/PyCQA/pycodestyle/issues/506

This proved simpler because I realized something important: Most of the
code that the old integration was using was in fact doing nothing of
value. Since we can't meaningfully allow users to use command-line
options as parameters to `python setup.py flake8`, we just remove the
work that we do to set up those attributes and parse them.

This allows us to add --append-to-select and --append-to-ignore at a
later date that will work as expected and in a consistent manner.

Extract the files changed in a particular commit or set of commits
for the commit and qrefresh mercurial hooks.

Previously, we forced Flake8 to be installed in whatever Python
environment that the user was using. This allows someone to use Flake8
from a virtual environment, e.g., if you use the tox.ini from this
commit, you can do:

    tox -e venv -- flake8 --install-hook git

And that will allow you to use the Python and Flake8 from
./.tox/venv/bin/python. This means that you can avoid installing
Flake8 globally and still have a working commit hook.