Remarks on best approximation in R-trees