Added support for loading and applying CatBoost models to data stored in ClickHouse.
Why it matters
This feature enables users to integrate CatBoost machine learning models directly within ClickHouse, allowing efficient scoring and inference on large datasets without exporting data to external environments. It simplifies workflows and accelerates data-driven decision making by leveraging ClickHouse's performance and CatBoost's predictive capabilities.How to use it
Users can load CatBoost models into ClickHouse and apply them to columns of data using new functions or table engines designed for this purpose. Typically, this involves importing a CatBoost model file and then invoking model prediction functions inline within SQL queries to score data directly. Refer to ClickHouse documentation and the PR for exact syntax.