Lower Complexity Bounds for Finite-Sum Convex-Concave Minimax Optimization Problems