Fix rendered newline escapes in GP regression notebooks

[jwwtc]

Motivation

• The GP notebooks contained literal escaped newline sequences (\\n) in cell source strings which caused GitHub to show raw \n text instead of rendered markdown/code.
• The goal was to make the notebooks display normally while preserving the existing model-loading examples.

Description

• Rewrote src/gaussian_process_regression/gp_regression_a.ipynb and src/gaussian_process_regression/gp_regression_b.ipynb so the cell source arrays contain real newline characters instead of escaped \\n sequences.
• Kept the notebook structure (cells, metadata.kernelspec, language_info, nbformat/nbformat_minor) and the model-loading code unchanged aside from newline encoding.
• Addressed the inline diff feedback by ensuring code and markdown sources are written as normal multi-line strings that render properly on GitHub.

Testing

• Parsed both notebooks with json.loads successfully to confirm valid JSON documents.
• Verified the first markdown and code cell source entries contain real newline characters using repr(nb['cells'][i]['source'][j]), which showed actual \n characters rather than \\n escapes.
• Confirmed no literal escaped-newline sequences remain by running rg '\\\\n' src/gaussian_process_regression/gp_regression_a.ipynb src/gaussian_process_regression/gp_regression_b.ipynb and observing no matches.

---

Codex Task