[FEA] uv pyproject.toml install support

Is your feature request related to a problem? Please describe.
The Python community in general is eager to adopt uv and pyproject.toml for libraries.

Describe the solution you'd like
Would love to be able to say uv add cuml (or other accelerated libraries).

Describe alternatives you've considered
If a single uv command isn't sufficient (due to alternate indexes and cu12/13 dependency groups), a reference pyproject.toml file would be appreciated.

I was exploring the pyproject.toml file at https://github.com/NVIDIA/dgx-spark-playbooks/tree/main/nvidia/portfolio-optimization/assets/setup and was very hopeful after a uv sync ran on my Spark DGX, however, when I went to run the notebook, I discovered that it didn't include Jupyter or cuML when running .

Additional context
The competition here on the Apple side is strong. I have a MBP and wanted to run the parakeet model in a Python script, all I needed to do was add this to the top of the script and run it with uv run. No Docker, no conda 30 minute installs, no slow pip...

#!/usr/bin/env python3
# /// script
# requires-python = ">=3.12"
# dependencies = [
#     "parakeet-mlx>=0.5.0",
#     "sounddevice>=0.4.6",
#     "numpy>=1.24.0",
#     "pyperclip>=1.8.2",
# ]
# ///

Also note that https://docs.rapids.ai/install/ doesn't include a single reference to Spark DGX. A simple note, call out would be appreciated to direct down the golden path.

Start with the RAPIDS install documentation at https://docs.rapids.ai/install/ and compare it with the referenced DGX Spark pyproject.toml setup. Determine whether the requested outcome is a documented uv workflow, a reference pyproject.toml, or a Spark DGX installation note. Done means the chosen path is documented with the required Jupyter and cuML details and a clear Spark DGX direction.