TODOs

• Try using introselect instead of quickselect for median selection in buildKdMap
• Add benchmark data for dynamic k-d trees
• Speed up all dynamic queries by using best-so-far answers from previous trees.
• Put all the linear benchmark/testing logic in one place (less copy-paste)