Implement some more similarity metrics

- [ ] Hamming distance: Linear algo. The number of bits that need to be changed to convert one vector into the other. Fast and useful for binary vectors
- [ ] Minkowski: Linear algo. It is a generalisation of Euclidean and some other distances like Manhattan i.e. by tuning some parameters, you get Euclidean/Manhattan or other distances
- [ ] Locality Sensitive Hashing: Nonlinear algo. Works by grouping vectors into buckets by processing each vector through a hash function that maximizes hashing collision as opposed to minimizing as is usual with hashing functions. Not suitable for large dimensionality vectors
- [x] Hierarchical navigable small world: Nonlinear algo. An adaptation of navigable small world (NSW) graphs where an NSW graph is a graph structure containing vertices connected by edges to their nearest neighbors.Good for high dimensionality data