Learning with Bandit Feedback in Potential Games