Generate an infohash reducing the dimentionality of vector coming in from PeARS

[nandajavarma]

KD-tree implementation that was initially expected to solve the search on multidimentional keys in a network was a failure since the tree cannot be updated online. If we use that datastructure, the idea is that the tree will be generated initially at the beginning with all the data points and then we can lookup easily afterwards. But we cannot add new data to that network again.

This means that, we have to stick with the PHT trie structure for now. As for the initial testing, we can try using a simple z-curve linearization algorithm on our vector.