A Setwise Approach for Effective and Highly Efficient Zero-shot Ranking with Large Language Models