ENH: Add min_samples_leaf parameter to fcluster to be able to set minimal number of samples in cluster

[dimadgo11]

Is your feature request related to a problem? Please describe.

Function scipy.cluster.hierarchy.fcluster(Z=Z, t=30, criterion='maxclust') may return clusters with very small sizes like 1, 2, 3 etc.
I believe that in many cases this behavior is not desired.

Describe the solution you'd like.

I expect min_samples_leaf or min_samples_per_cluster parameter to work this way: if it's set, then we walk through the dendrogram/tree/linkage matrix as usual, but stop splitting the node if number of samples in leaf is less then it is set. Then we freeze this branch and keep walking down the tree in other branches unless we get required number of clusters or hit this limitation.

Describe alternatives you've considered.

No response

Additional context (e.g. screenshots, GIFs)

No response