Add Colab notebooks, README updates, and Bayesian bug fixes

[Mattral]

Summary

This PR delivers the notebook integration, README polish, and several bug fixes found during notebook development.

New Content

• Added five fully executed Jupyter notebooks under notebooks/ (supervised, unsupervised, bayesian, reinforcement, neural networks).
• Updated README:
  • Removed broken CI badge.
  • Added Colab badge row.
  • Added dedicated "Try it — Colab notebooks" section with direct links.
  • Updated test counts and algorithm table rows.

Bug Fixes

• BayesianNetwork.query(): fixed axis-shrinkage logic for multiple evidence variables.
• BayesianNeuralNetwork: clamped log_sigma to [-8, 2] with gradient clipping to prevent overflow.
• GaussianProcessRegressor.predict(): corrected 1-D input handling to avoid kernel matrix mismatch.

Documentation

• Added notebooks entry to CHANGELOG.
• Updated roadmap and README to reflect new test counts and notebook availability.

Verification

• All five notebooks execute end-to-end with zero errors.
• 15 previously failing tests now pass; 11 remaining failures are unrelated NumPy 2.x drift.
• All examples run successfully.

This PR makes the notebooks accessible via Colab and ensures the Bayesian methods are robust and correctly tested.

[dagshub]

Join the discussion on DagsHub!

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB