Add general XGBoost demo notebook

[dantegd]

Adds an initial version of a notebook demonstrating GPU-accelerated XGBoost training with RAPIDS. The notebook walks through simulating/loading tabular data, configuring XGBoost for GPU execution using device='cuda', and training a binary classification model.

Notes

• Uses XGBoost 3.1 compatible parameters
• Requires CUDA-capable GPU

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[csadorf]

I'm in favor of running this within our CI, but I think it would be confusing to users that we have an XGBoost demo notebook within our notebooks/ directory that is generally documented as providing examples for cuML. Can we place the notebook somewhere else or maybe put it into a (hidden) sub-directory that provides a brief explanation as to why we include this with the repository?

I think placing it somewhere within the ci/ directory might be more appropriate. We can easily adjust the ci/test_notebooks.sh directory to deal with that, e.g., by explicitly listing it or just moving/linking it into the notebooks/ directory right before we execute the tests.

[trivialfis]

Please consider using the quantile DMatrix instead

[dantegd]

/merge