Add a side hierarchic nodal soln implementation

[lindsayad]

Given the previous comment here (which I've temporarily kept), I anticipate some debate on this, although I think the implementation is defensible given what we already do for projecting discontinuous solutions onto nodes for other output cases

[lindsayad]

If we agree on this implementation, I'd probably go and temporarily skip the affected MOOSE tests

[lindsayad]

This is how side_hiearchic_out.e in MOOSE looks like on QUAD9 with this PR visualized via paraview. My current choice to not encode a value for element interior nodes is obvious. What do people think? I'll note that previously this gold file just had all zeroes for the aux_side_var and didn't even have any representation of side_var due to some quirks in the MOOSE output system. I'm a little surprised that @roystgnr decided this Exodiff test fulfilled this stated requirement, but I've certainly done worse in my time I'm sure.

The system shall initialize SIDE_HIERARCHIC variables as instructed

With this PR: CONSTANT SIDE_HIERARCHIC side_var

With this PR: FIRST SIDE_HIERARCHIC aux_side_var

[moosebuild]

Job Coverage, step Generate coverage on f68ea33 wanted to post the following:

Coverage

	125e14	#4494 f68ea3
	Total	Total	+/-	New
Rate	65.65%	65.65%	+0.00%	100.00%
Hits	78738	78751	+13	15
Misses	41207	41207	-	0

Diff coverage report

Full coverage report

This comment will be updated on new commits.