Distribute the PURL dataset as a separate package

It would be great to distribute the PURL validator datasets as separate packages. This would help faster updates. This would be obviously a wheel for Python, but we could also consider similar approaches for Go and Rust (with some caveats: native Go and Rust do not support distributing pre-built binaries) .... other ecosystems like Conda/Pixi may like to bundle these as separate packages. The rationale is that:
- the code to read/expose the PURL Fsts is mostly stable
- the data, e.g., the list of PURLs changes hourly (but practically a daily or weekly snapshot may be enough)