Segment Any Point Cloud Sequences by Distilling Vision Foundation Models