Refactor with UV, swap docs to jupyterbook, add nilearn nv download integration

[ejolly]

• Switch to pyproject.toml managed by uv
• Switch docs to jupyter-book
• Setup new GA workflow to use uv to test, build, docs, and publish
• Fix deprecation code warnings

[ejolly]

@ljchang @nirjacoby Can you take a quick look at this by cloning the branch and seeing if you run into any odd issues using this version? All tests pass and I've switched our docs to jupyter-book

The README contains the uv commands to run tests and build docs locally. I can merge this into master and create a new release to unblock issues related to pinned versions.

I'm currently remaking some tutorials and cleaning up the docs:

• Basic Brain_Data tutorial
• Basic Design_Matrix tutorial
• Basic Adjacency tutorial
• Basic full workflow tutorial
• Advanced tutorials (refined versions of current)
• Integrate commits from WIP: Add more MNI templates #423

Addresses:

• numpy locked to old version #445
• cannot import name '_get_dataset_dir' from 'nilearn.datasets.utils' (/opt/conda/lib/python3.11/site-packages/nilearn/datasets/utils.py) #444
• Move documentation to jupyter book #433
• Need to update neurovault file downloading with new nilearn API #429
• Known compatibility issue Python 3.11 and joblib #430
• fetch_pain() issue #422
• Adjust sigma estimation for standard error computation in regress #287 (suppress warnings)

[ejolly]

Closing — superseded by #453, which carries the UV migration, the JupyterBook docs swap, and the nilearn/neurovault download integration into the 0.6.0 rewrite along with the broader refactor.