Create descriptor of every clip. Takes clusters and assigns each cuboid to its type. Next, the cuboid types are histogrammed to create a descriptor for each clip in each dataset. INPUTS DATASETS - array of structs, should have the fields: .desc - length N cell vector of cuboid descriptors .ncilps - N: number of clips clusters - cuboid clusters csigma - soft assign see clipdesc OUTPUTS data - length N cell vector of (nclips x p) arrays of data