Implement K-Means Clustering

Given a list of n points in d dimensions as points, an integer k, a maximum number of iterations max_iters, a tolerance tol, and an optional random seed seed, implement K-means clustering from scratch. Return the final centroids, the cluster assignment for each point, and the number of iterations performed. Use Euclidean distance, initialize centroids by sampling k distinct points, and stop when centroid movement is at most tol or when max_iters is reached. If a cluster becomes empty, reinitialize its centroid to the point farthest from its currently assigned centroid.