[CI] Docs build fails due to Figshare network error in getting_started.ipynb

[csadorf]

Summary

Docs build fails when executing the cuml-accel/examples/getting_started.ipynb notebook. The notebook attempts to download the Covertype dataset via sklearn.datasets.fetch_covtype(), which fetches data from Figshare. The download fails with an HTTPError.

In #7581, we switched from downloading directly from the UCI dataset repository (which was down) to using sklearn's fetch_covtype utility, which was supposed to use a more reliable mirror. However, sklearn's mirror (Figshare) is now also experiencing intermittent availability issues in CI environments.

This is a known community issue as documented in scikit-learn/scikit-learn#30761 .

Failing test/component: docs-build (nbsphinx notebook execution)

Failure observed in:

• https://github.com/rapidsai/cuml/actions/runs/59135054303

Environment

n/a

Test Details

• Test file: docs/source/cuml-accel/examples/getting_started.ipynb
• Test name: Notebook cell execution during Sphinx docs build
• Error message:

  nbsphinx.NotebookError: CellExecutionError in cuml-accel/examples/getting_started.ipynb:
  ------------------
  dataset = fetch_covtype(as_frame=True)
  X = dataset.data
  y = dataset.target
  
  print(f"n_samples = {X.shape[0]}, n_features = {X.shape[1]}")
  ------------------
  
  ----- stderr -----
  /opt/conda/envs/docs/lib/python3.13/site-packages/sklearn/datasets/_base.py:1519: 
  UserWarning: Retry downloading from url: https://ndownloader.figshare.com/files/5976039
  ------------------
  
  HTTPError: HTTP Error (download from Figshare failed)