Low-Rank Inducing Norms with Optimality Interpretations