Text this: K-means and agglomerative clustering for source-load mapping in distributed district heating planning