Avoid tiny numpy arrays in numba: create numba helper routines file #1680

Sub-issue of #1648 to track: should create a file like numba_math.py with helper routines acting on tuples.

As mentioned in the original issue, the idea is to put helper functions here, e.g. _numba_norm3(ux, uy, uz) which allows value = _numba_norm3(ux, uy, uz) instead of value = (ux**2 + uy**2 + uz**2)**0.5. These helper functions will be used throughout various regions of the code while addressing #1648.

Although, after looking into it a bit, I have learned that using fixed-length tuples such as u = (ux, uy, uz) is just as efficient as using individual components directly; the inefficiencies come into play if you ever create a list or numpy array. So, I believe helper functions working on tuples would be better, e.g. _numba_norm3(u) to handle value = (u[0]**2 + u[1]**2 + u[2]**2)**0.5.