Lucid.CLIP.CompLabelingMainPerSlice

This task type uses CLIP to do component labeling per slice, and only keep the main object (with the largest cross-section) per slice.