create a similarity matrix for the clustering notebook

[stevehadd]

calculate the distance in feature space between each of the pairs of the cluster centres and display as a matrix (colour coded, so white where similar and red where very different