One Tree, Many Cuts

A gardener has measured spots in a field where seedlings came up, as [x, y] in metres. Hierarchical clustering builds one tree of merges, and you choose afterwards how many groups to cut it into.

Task: write cut_sizes(points, levels).

• For each number in levels, in the given order, fit AgglomerativeClustering(n_clusters=n) on the points as given. Keep the default linkage and don't scale the points.
• Count how many points are in each cluster.
• Return one list per level, each sorted from largest to smallest, in the same order as levels.