Lisa wants these two methods implemented
input: h x w x z x 1 with segments
2.2 is already implemented in https://github.com/Kainmueller-Lab/brainbow/blob/main/supervoxelize.py#L53C1-L54C1
image: RGB image
sv_label: watershed supervoxels with labels
sv_image: supervoxels filled with mean color from RGB image
(2.3) split supervoxel with different colors
--> not implemented yet
(2.4) demix/merge supervoxel
--> not implemented yet