Queueing Matching Bandits with Preference Feedback