Doubly Optimal No-Regret Online Learning in Strongly Monotone Games with Bandit Feedback