Bandits With Preference Feedback A Stackelberg Game Perspective