Full batch gradient descent logistic regression from scratch

Implement vectorized training loop with batch gradient descent using NumPy only.
Plot loss across iterations and report final classification metrics on test split.

Start by locating the repository's Jupyter notebook and the current model or training entry point. Implement the requested NumPy-only full-batch training flow, then confirm that loss is plotted across iterations and final classification metrics are reported on the test split.

Written by the indexing model from the issue text.