zenseact/zod ? reverse-engineered prompt

Build me a Python toolkit for working with the Zenseact Open Dataset.

I want a small but useful library that lets me download the dataset, browse it, and use it in machine learning projects. It should handle the three dataset types, Frames, Sequences, and Drives, and make it easy to load data, inspect annotations, and work with the different sensor streams. Please include a command line tool too, so I can download a dataset from a link, choose what to include, and turn on or off things like images, lidar, vehicle data, and anonymized versions.

I also want a few example notebooks or sample scripts that show how to use the dataset and how to train a simple object detection model with the data. Make the install story simple for basic use, for CLI use, and for the full setup with all extras. If you need to check current docs or best practices for the Python packaging pieces, look them up online first.