What this PR does / why we need it:

Users onboarding to a Feast have no standardized way to explore their project or learn how to retrieve features. They must manually read documentation and write boilerplate code to understand what entities, feature views, and services exist and how to use them for training or serving.

This PR adds a new copy_demo_notebooks() function to the Feast Python SDK (and a corresponding feast demo-notebooks CLI command) that generates tailored Jupyter notebooks for any Feast project. The notebooks auto-adapt to the project's feature_store.yaml configuration and provide ready-to-run examples for data scientists.

A single command generates a set of project-specific demo notebooks:

from feast import copy_demo_notebooks
copy_demo_notebooks()

Or via CLI:

# feast demo-notebooks

Generated notebooks:

01_feature_store_overview.ipynb - Explore registered entities, feature views, services, and data sources
02_historical_features_training.ipynb - Build training datasets using point-in-time correct historical retrieval
03_online_features_serving.ipynb - Materialize features and serve them at low latency for inference