Was disallowing adjacent loops to be split which is correct for a single split across a face, but not fore BM_face_split_n