Doubly Optimal No-Regret Learning in Monotone Games