[pyclustering.cluster.silhouette] Distance matrix support

[annoviko]

Introduction
There is request to have a distance matrix as an input parameter for Silhouette method.

When trying to determine the optimal number of medoids, I will want to use the silhouette functionality. Can you confirm that it is correct to pass the distance matrix into this and not the original dataframe? I am presuming that I can't pass a dataframe with non-numeric features anyway.

Description
Silhouette should be able to calculate amount of clusters using distance matrix as an input data in the same manner as K-Means, K-Medoids, etc.