Am I correct in saying that phenix.real_space_refine does not (yet?) have NCS restraints (that I need for refining a molecule with broken symmetry)?