The page was removed along with the extension in 1127013.

Also ensure any future MkDocs warnings cause the build to fail in the tests
so we don't introduce similar errors in the future.

* Install deps.
* Ensure test fails if deps are missing.
* Update dictionary for recent docs changes.

Fixes #601. Merged in 6f87b32 from the md3 branch and did a lot of cleanup.

Changes include:

* Removed old docs build tool, templates, etc.

* Added MkDocs config file, etc.

* filename.txt => filename.md

* pythonhost.org/Markdown => Python-Markdown.github.io

* Markdown lint and other cleanup.

* Automate pages deployment in makefile with `mkdocs gh-deploy`

    Assumes a git remote is set up named "pages". Do

        git remote add pages https://github.com/Python-Markdown/Python-Markdown.github.io.git

    ... before running `make deploy` the first time.

Not sure this is the best way to go, but it works. I'm not crazy about
running the spellcheck against the built docs, but aspell has a builtin
option to easily ignore everything in `<code>` tags which greatly simplfies
things.

I looked at Doug Hellmans' sphinxcontrib-spelling package which does
something similar for Sphinx. However, as Sphinx uses rST and the rST
parser outputs a parse tree, Doug is essentially taking that parse tree
and running the spellcheck on the appropriate parts (skipping code, etc.).
He did a nice [writeup][5] of his development process if you are interested.

As Python-Markdown's parse tree is represented as HTML (through ElementTree)
I would have to use HTML anyway. And [PyEnchant][2] doesn't currently have
good support for HTML. So I used [aspell][3], with inspiration from the
[git-spell-check][4] hook.

[1]: http://sphinxcontrib-spelling.readthedocs.org/en/latest/index.html
[2]: https://pythonhosted.org/pyenchant/
[3]: http://aspell.net/
[4]: https://github.com/mprpic/git-spell-check
[5]: http://doughellmann.com/2011/05/26/creating-a-spelling-checker-for-restructuredtext-documents.html