Consider a range counting query, where the query asks for the number of points inside a range. Assume that you have a dataset of 2-d points and you want to create an index based on kd-trees to answer efficiently range counting queries. Explain how you will modify the original kd-tree data structure and give an algorithm that will use this modified kd-tree to answer efficiently range counting queries. What is the worst case performance of your data structure for range counting queries?