Enhancing Privacy in Decentralized Min-Max Optimization: A Differentially Private Approach