Improved Regret for Bandit Convex Optimization with Delayed Feedback Y uanyu Wan 1,2,3, Chang Y ao