Add clustering pipelines

Should include a suitable baseline clustering pipeline. One option for a good baseline would be randomly splitting the data into two clusters, or n clusters if the user supplies the expected number of clusters.