Exploiting hierarchical clustering for finding bounded diameter minimum spanning trees on euclidean instances