Session 10C - Coresets for Clustering in Euclidean Spaces: Importance Sampling is Nearly Optimal