This basically commits the transpiled CoffeeScript JS (with minor
cleanup) and removes coffee build support.

A tiny amount of support for xblocks exists, because external users
may have xblocks with coffee. But no coffee in our tree anyway.

Covers some of the basics of how our Sass is compiled in edx-platform.

- Remove old docs
- Move testing.rst
- Update READMEs and index.rst for docs.

This makes it possible to refer to multiple book from a single app.

This also means that Open edX branches will be able to pass bok-choy
tests (at least if the books have already been built.)

TNL-6743

This test lives in `common/test/acceptance/tests/lms/test_lms_dashboard.py`, so the paver string should be `paver test_a11y -t lms/test_lms_d...`

This commit contains back end changes necessary to load programs from the catalog in all contexts. The existing program munging utility is applied as late as possible to avoid conflating this work with changes to the front end; those will be made separately.

ECOM-4422

ECOM-5936

Rendering model states when running migrate in Django 1.8 consumes
on the order of 4x-6x more memory than <1.8.

References:
* https://code.djangoproject.com/ticket/24745
* https://code.djangoproject.com/ticket/24743

This is causing failures in Jenkins as the workers run out of memory
and either processes cannot fork or the Out Of Memory Killer
kills a process that is running tests.

Disabling migrations until we do either or both of these:
* Upgrade to Django 1.9 and turn them back on. See EV-92 "Discovery: Upgrade edx-platform to Django 1.9"
* Address TE-1310 "Unit tests should be able to run without migrations"

Adds a context-sensitive link to learner documentation from the LMS. The base URL defaults to Open edX learner doc and can be overridden with the base URL for edx.org learner doc.  Also relabels the tab for support from help to support.

Fixed import failures that were preventing the platform API documentation from building. The problem is that the doc building environment doesn't need to install all the Python modules that the platform software does, but when Sphinx tries to import select modules so that it can use autoclass to publish docstrings it tries to import all the modules that are imported by those modules, recursively. I mocked the missing modules.