Lifted polymatroid inequalities for mean-risk optimization with indicator variables