Add in exact radii and normalized radii methods

[lcgraham]

Use scipy.spatial.Delaunay to determine neighboring cells. Then determine distance to the furthest neighbor. See bet.calculateP.calculateError for hints/guidance. Make sure that we use radii as a cutoff in query for local volume emulation.

[lcgraham]

Use kdtree to find/do the connectivity to estimate/determine neighboring cells.

1. For all emulated points in cell get the 2 nearest neighbors.
2. The set of second nearest neighbors for all emulated points included in a cell are the neighboring cells. Then we calculate the distances/radii exactly from this. (The might be possible without the emulated points.) This is an approximate connectivity.

@smattis This is a summary from yesterday's discussion and might impact the calculateError sub package.

[lcgraham]

Needs more work to take into account bounded domain...