This PR implements the BallTree data structure from petal-neighbors as requested in issue #1, providing efficient nearest neighbor search capabilities to the pypetal project.

Closes #1

Summary

I've implemented the BallTree data structure from petal-neighbors 0.18.0 for the pypetal project. Here are the changes:

Files Modified

1. Cargo.toml - Added petal-neighbors = "0.18" dependency and renamed the library from decomposition to pypetal to serve as the main module

2. src/lib.rs - Updated to expose both decomposition and neighbors submodules through the main pypetal Python module

3. src/decomposition.rs - Made the decomposition function public for use from lib.rs

4. setup.py - Updated to use pypetal.pypetal as the Rust extension name

Files Created

5. src/neighbors.rs - New module containing the BallTree wrapper with:
   • Constructor: BallTree(points) - Creates a ball tree from a 2D numpy array using Euclidean distance
   • Properties: n_samples, n_nodes
   • Methods:
     • query_nearest(point) - Returns the nearest neighbor index and distance
     • query(point, k) - Returns k nearest neighbor indices and distances
     • query_radius(point, radius) - Returns all neighbor indices within the given radius
   • Unit tests: 4 comprehensive tests for construction and all query methods

Python Usage Example

import numpy as np
from pypetal.pypetal import neighbors

# Create sample points
points = np.array([[0.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 1.0]])

# Build the ball tree
tree = neighbors.BallTree(points)

# Query nearest neighbor
idx, dist = tree.query_nearest(np.array([0.1, 0.1]))

# Query k nearest neighbors
indices, distances = tree.query(np.array([0.0, 0.0]), k=2)

# Query all neighbors within radius
indices = tree.query_radius(np.array([0.0, 0.0]), 1.5)

All tests pass and clippy reports no warnings.