Remove direct dependencies on numpy/scipy in edx-platform

[ormsbee]

We still use them in openedx-calc (for now) and openedx-chem (for now), but there are places in edx-platform where we can substitute our current usage with Python stdlib calls, such as:

edx-platform/cms/djangoapps/contentstore/api/views/course_quality.py

Lines 270 to 276 in 25536bb

	return dict(
	min=min(data),
	max=max(data),
	mean=np.around(np.mean(data)),
	median=np.around(np.median(data)),
	mode=stats.mode(data, axis=None)[0],
	)

The long term goal would be to isolate numpy/scipy to just the parts of the XBlocks that need them (and possibly isolate them to just the codejailed execution environment). This would reduce memory usage and make upgrades less painful.

We have to be really careful if we're touching anything related to grading.