pola-rs/polars ? reverse-engineered prompt

Build me a really fast DataFrame query engine in Rust, with a friendly Python API on top, like Polars.

I want it to feel easy to use for data work, with both eager and lazy query styles, plus simple filtering, grouping, sorting, joining, and aggregations. It should be optimized for speed, use multiple cores well, and handle big datasets without needing everything in memory if possible. If it makes sense, include support for reading common file formats like Parquet and make the API expressive enough to chain operations naturally.

Please set up the Rust core first, then the Python bindings, and make sure there are examples and basic docs so I can try it quickly. If you need to look up current docs or best practices for Rust or Python bindings, go ahead and do that. I mostly want a clean working version I can run, test, and extend later.