Chdb Datastore

The main pitch is you can swap `import pandas as pd` for `import chdb.datastore as pd` and your existing code gets faster through lazy execution and SQL compilation. Where it really shines is cross-source joins: pull a table from MySQL, join it with a parquet file, aggregate, done. Supports 16+ connectors including S3, PostgreSQL, MongoDB, Delta Lake. The skill covers the full pandas-compatible API (209 methods), connection patterns for different sources, and when to use this versus raw SQL queries. If you're hitting pandas performance walls or need to join data across databases and files without ETL, this is the move.