Use uv for dependency management

[kylebarron]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

Right now this project uses a strange mix of pip, pip-compile, and conda. This means dependencies have to be declared in several places, and pip-compile gives hard-coded environment files per Python version, which makes it hard to understand where those lockfiles came from and how to update them.

I'd like to suggest that this project considers uv for Python package management. I'd argue it's the modern standard for package management. It's fast and widely used.

The only reason I can think of to use conda instead for developing would be if we needed GPU support in development. But aside from having a cuda-enabled environment.yml file, there's no mention of CUDA in the repo. This linked example seems to be missing from the repo:

datafusion-python/examples/README.md

Line 54 in 79c22d6

- [Executing SQL on cuDF](./sql-on-cudf.py)

If we did need to stay in the Conda ecosystem, I'd highly recommend pixi, which manages native dependencies but provides a full platform-independent lockfile across multiple Python versions.

Describe the solution you'd like

Use uv.

Describe alternatives you've considered

Status quo.

Additional context
Add any other context or screenshots about the feature request here.

[ion-elgreco]

UV is a blast to use