(2) 函数说明：
% this function implements different types of evaluation metrics of multi-label learning
% please see NIPS 2011 "A Literature Survey on Algorithms for Multi-label Learning" for detailed definitions of all metrics
% Input:
% label_matrix_prediction, m x n matrix
% label_matrix_gt, m x n matrix, each entry \in {-1, 1}
% options, has the following fields
% -- type, 'ranking', 'example-based-partition', 'label-based-partition'