Feature description

Description

The current implementation of the K-Nearest Neighbors (KNN) algorithm in machine_learning/k_nearest_neighbors/ uses the Euclidean Distance by default.

To enhance the algorithm's functionality and completeness, this issue proposes adding two additional distance metrics that are commonly used in the KNN algorithm:

1. Manhattan Distance (L1 norm)
2. Minkowski Distance (Lp norm), which is a generalized form that includes both Euclidean (p=2) and Manhattan (p=1) as special cases.

Proposed changes

1. Modify the KNN implementation (k_nearest_neighbors.py or similar file) to include a new parameter, perhaps distance_metric (e.g., set to 'euclidean' by default), that accepts 'manhattan' and 'minkowski' as values.
2. Implement the calculation for the Manhattan Distance.
3. Implement the calculation for the Minkowski Distance, using an additional parameter p for the power (e.g., defaulting to p=2 if distance_metric is 'minkowski').
4. Add comprehensive unit tests to ensure that the algorithm performs correctly when using these new distance metrics.