Normal normals

[roystgnr]

A user suggested that they might have seen normal vectors inverted on a flipped Tri element. I couldn't reproduce that, but I did see normal vectors inverted on a flipped Edge element. This PR fixes that, and adds more unit tests to make sure such a gross error shouldn't recur. More subtle potential errors are going to need more subtle tests at some point.

[moosebuild]

Job Coverage, step Generate coverage on b0cd68f wanted to post the following:

Coverage

	db32ea	#3971 b0cd68
	Total	Total	+/-	New
Rate	62.33%	62.33%	+0.00%	100.00%
Hits	72643	72646	+3	6
Misses	43901	43902	+1	0

Diff coverage report

Full coverage report

This comment will be updated on new commits.

[roystgnr]

Rebased on master to see if the MOOSE fixes have made it in yet.

[roystgnr]

And again

[jwpeterson]

This looks lie a pretty self-contained bug fix, so probably a candidate for the 1.8.x branch @roystgnr?

[roystgnr]

Yeah, I'd say so. Our only problem with it was the downstream app code that had been relying on the broken behavior.