Estimate dof_vel by 2 quaternions

https://github.com/nv-tlabs/ASE/blob/main/ase/utils/motion_lib.py#L332
When estimate dof_vel with def _local_rotation_to_dof_vel(self, local_rot0, local_rot1, dt)
why it is diff_quat_data = quat_mul_norm(quat_inverse(local_rot0), local_rot1) instead of
diff_quat_data = quat_mul_norm(local_rot1, quat_inverse(local_rot0))?
The later one seems to be more accurate.