Interpolate Point Cloud to Regular Grid consumes massive memory

[imikejackson]

This filter essentially caches a neighborlist at each voxel location. If there are a lot of points that can lead to massive amounts of memory that is needed to store these lists.

Based on a pipeline that using this filter, another approach may be needed for this filter. Such as computing the statistics "on the fly" so we don't have to store the neighbor list. Or computing the full list + statistics in sections along an axis so we only have to hold part of the image geometries neighborlists in memory.

Possibly compute the approximate time left
parallel algorithm possible?
Additional progress feedback at the start and the end of the algorithm to let the user know something is happening.
Add a cancel check into the algorithm