Improve Soma Skeletonizations

[william-silversmith]

We could conceivably add a third pass to the skeleton construction by using much higher mip levels to find the somata with a large enough context that it's nbd. We would then find a way to merge the resulting good skeletons with the mess that results from many smaller fields of view. @jabae had some thoughts about this.

Problems

• avocados (use fill holes and fix_avocados in Kimimaro)
• selecting the right mip level
• selecting the right stride
• merging the new skeletons with the old ones

Fixing the pinky100 skeletonization would be a great target.

[william-silversmith]

For certain simple approaches such as drawing lines from a point, it might be possible to dispense with chunking and just use the 256x256x320 resolution data. However, to be more precise (such as drawing skeletons from the center) we'll need higher res data (maybe 64x64x40 or 64x64x80) to see medium thickness processes.

[william-silversmith]

max physical displacement between mip levels = sqrt(2) * (2^mip - 0.5) * r