Scbsp

Implemented generator-based approaches in several functions to reduce memory usage.

This release simplifies the dependencies and optimizes matrix multiplication operations.

This release simplifies the dependencies and optimizes matrix multiplication operations.

New Features:
- Introduced an optional GPU mode to enhance processing speed and efficiency.

New Features:
- DataFrame Output: The function has been updated to return a Pandas DataFrame, enhancing the structure and accessibility of the output data. The DataFrame contains two columns: `gene_names` and `p_values`, providing a clear and comprehensive overview of the gene expression analysis.

New Features:
- Enhanced Flexibility in Distance Calculation: With this release, scBSP users gain the ability to select between two powerful algorithms for distance calculation:
- Ball Tree is now the default distance calculation method, optimized for efficiency and accuracy in processing biological data.
- HNSW (Hierarchical Navigable Small World graphs) has been introduced as an optional method, providing an alternative that balances speed and precision for large datasets.