allow student state of module other than problem to be reset by instructor

Adding in reference and UI for Tender discussion submission widget

Studio - JQUI's default date picker z-index bug

list xml_attributes (new field since xblock integration) as system metad...

optimize db round trips, by getting the course descriptor outside of the...

when grading, check for always-regrade at the section level too

modify AssignmentFormatGrader to act like a SingleSectionGrader on single sections

Fix/will/chem customresponse bug

context for scripts to access.
This was removed accidentally during an earlier refactoring,
which caused customresponse to raise an exception for chem problems.
Unit tests verify the changes

Fix/will/random seed not saved

Enable dev env support for sass source maps

so that the same problem loads when the user
checks/saves/resets

create-dev-env.sh is failing because rednose 0.3.3 seems to have
fallen off the face of the Earth. 0.3 is still available so reverting
to that.

Enable executing the lms jasmine tests as part of the build process

Fix/vik/oe and progress

Capa problem state was not being copied to module state
during module initialization.
This meant that when check_problem was called the first time,
the seed would be re-generated and saved.

list xml_attributes (new field since xblock integration) as system metadata which should not be edited in Studio

studio - resolving a bug where JQUI sets the z-index of our datepickers to 1, which is less than our dynamic tooltips, causing visual display/access issues.

Fix/will/bug 288

change submission history to be ordered by id

was not set to done before resetting it.
To avoid to small chance that we'll get the same seed when
we reset, try to get a new seed multiple times.

this makes it possible to reset open ended response problems

HOTFIX: Accept either a list of possible values, or a string as a value for

I'm pulling this in myself because it's urgent, it's the weekend, and I don't want this to get forgotten.

comparison of correctness in multiple choice.

Multiple choice code is scattered and sometimes sends a list of
choices for the value, and sometimes a single string. We used to use
"in" which scarily handled both cases (list or substring search),
but that caused a bug when you had two choices like choice_1 and
choice10. Moving to == caused us to break when lists were sent to
us. So this ugly code is extra paranoid and checks both possibilities.

This really needs a better cleanup.