Refactor code for 3d Nédélec shape functions

[nmnobre]

Pending benchmarking...

[roystgnr]

I certainly like this, and I doubt it'll show any problems in benchmarking. I'd bet you'll need to run something that really hammers assembly alone, on a large mesh, or any change in timings will end up swamped by overhead and/or the solver.

[moosebuild]

Job Coverage on 8f16abc wanted to post the following:

Coverage

	61691f	#3874 8f16ab
	Total	Total	+/-	New
Rate	62.72%	62.74%	+0.02%	86.52%
Hits	69377	69298	-79	77
Misses	41234	41157	-77	12

Diff coverage report

Full coverage report

Warnings

• New new line coverage rate 86.52% is less than the suggested 90.0%

This comment will be updated on new commits.

[nmnobre]

Errrmmm, so I compiled libMesh with MOOSE's default configuration plus --enable-perflog.
Then ran the 3d Nédélec one example, i.e. vector_fe/ex4, with grid_size = 15, 5 times with
for i in {1..5}; do ./example-opt element_type=TET14 -pc_type jacobi | grep "| FE " -A2; done.

BEFORE:

| FE                                                                                                              |
|   compute_shape_functions()        334800     0.6720      0.000002    0.6720      0.000002    7.85     7.85     |
|   init_shape_functions()           334800     0.6295      0.000002    0.6295      0.000002    7.35     7.35     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.6651      0.000002    0.6651      0.000002    7.88     7.88     |
|   init_shape_functions()           334800     0.6184      0.000002    0.6184      0.000002    7.32     7.32     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.6665      0.000002    0.6665      0.000002    7.85     7.85     |
|   init_shape_functions()           334800     0.6228      0.000002    0.6228      0.000002    7.34     7.34     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.6638      0.000002    0.6638      0.000002    7.81     7.81     |
|   init_shape_functions()           334800     0.6203      0.000002    0.6203      0.000002    7.29     7.29     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.6720      0.000002    0.6720      0.000002    7.89     7.89     |
|   init_shape_functions()           334800     0.6219      0.000002    0.6219      0.000002    7.30     7.30     |

AFTER:

| FE                                                                                                              |
|   compute_shape_functions()        334800     0.7063      0.000002    0.7063      0.000002    8.00     8.00     |
|   init_shape_functions()           334800     0.8199      0.000002    0.8199      0.000002    9.28     9.28     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.7121      0.000002    0.7121      0.000002    8.01     8.01     |
|   init_shape_functions()           334800     0.8253      0.000002    0.8253      0.000002    9.28     9.28     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.7071      0.000002    0.7071      0.000002    7.96     7.96     |
|   init_shape_functions()           334800     0.8180      0.000002    0.8180      0.000002    9.21     9.21     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.7095      0.000002    0.7095      0.000002    8.01     8.01     |
|   init_shape_functions()           334800     0.8161      0.000002    0.8161      0.000002    9.21     9.21     |
| FE                                                                                                              |
|   compute_shape_functions()        334800     0.7188      0.000002    0.7188      0.000002    8.09     8.09     |
|   init_shape_functions()           334800     0.8235      0.000002    0.8235      0.000002    9.27     9.27     |

What's the difference between compute_shape_functions() and init_shape_functions()?
Here, init is about 32% slower and compute is 6% slower...