[Dev] Update News in Readme

[LeiWang1999]

This pull request introduces support for INT4 data types across various components of the BitBLAS library, including matrix multiplication and dequantization operations. The changes enhance the library's ability to handle mixed-precision operations and improve performance for certain data formats.

Enhancements and New Features:

• INT4 Support in Matrix Multiplication:

  • Added support for INT4 and UINT4 data types in bitblas/ops/general_matmul/__init__.py and updated related configurations and initializations. [1] [2] [3] [4]
  • Introduced new schedulers for INT4 data types in bitblas/ops/general_matmul/tilelang/dense/__init__.py and bitblas/ops/general_matmul/tilelang/dequantize/__init__.py. [1] [2] [3] [4] [5]

• Performance Optimization:

  • Adjusted local memory allocation sizes in bitblas/ops/general_matmul/tilelang/dense/matmul_tensorcore.py to optimize performance for INT4 and INT8 data types. [1] [2] [3] [4] [5] [6] [7]

Codebase Maintenance:

• Code Cleanup:
  • Removed an outdated comment in bitblas/gpu/intrin/lop3.py related to uint4 subtraction.
  • Updated data type mappings and assertions to include INT4 in bitblas/gpu/intrin/lop3.py. [1] [2] [3] [4] [5]

Documentation Updates:

• README Enhancements:
  • Added latest news about high-performance INT4 operations and updated the support matrix to include INT4 in README.md. [1] [2]

These changes collectively enhance the functionality and performance of the BitBLAS library, particularly in handling INT4 data types, and ensure that the documentation is up-to-date with the latest features.