fix: pin sklearn for compat with model pickles

[ceholden]

Description

This PR pins the scikit-learn package to the same major/minor version used by the model pickles. While I've had no issue running this with scikit-learn==1.8.0, it's risky to use different versions than what the pickle was saved as. Pickling models is an easy but complicated format for distributing trained models, and breakage across versions is one of the most common issues. A format like ONNX would be a better option for the future, but for now this PR locks the versions to stay in sync.

This resolves the warning emitted:

>>> loading lightgbm as tune machine learning model
/<path to python env>/lib/python3.13/site-packages/sklearn/base.py:463: InconsistentVersionWarning: Trying to unpickle estimator LabelEncoder from version 1.1.2 when using version 1.8.0. This might lead to breaking code or invalid results. Use at your own risk. For more info please refer to:
https://scikit-learn.org/stable/model_persistence.html#security-maintainability-limitations