Near Optimal Sample Complexity Bounds for Learning Latent k-polytopes and applications to Ad-Mixtures