fix: compatible with PEP 440

[zhyncs]

Motivation

Modifications

Checklist

• Format your code according to the Contributor Guide.
• Add unit tests as outlined in the Contributor Guide.
• Update documentation as needed, including docstrings or example tutorials.

[zhyncs]

twine check dist/*
Checking dist/sgl_kernel-0.0.2.post0+cu124-cp310-cp310-manylinux2014_x86_64.whl: PASSED

[zhyncs]

@yizhang2077 Due to PyPI's limitations, it's not possible to differentiate between CUDA versions. I plan to update only the CUDA 12.1 version on PyPI, while other versions like 11.8 and 12.4 can be updated in GitHub releases (especially 11.8 @ispobock please follow up). There's no need for a separate compilation for each Python version, compatibility with Python 3.9+ is sufficient, so there will ultimately be just one wheel package on PyPI.

[zhyncs]

update:
Slightly different from the previous description, to simplify, multiple Python version whl files are finally provided. I can later merge them into one compatible with Python 3.8+. CUDA 12.1 is used by default, which is fine for overseas users, but many users in mainland China may still be using CUDA 11.8.

https://pypi.org/project/sgl-kernel/#files