Limbo-ml

* Fix problems with the documentation build.

* Documented additional camera metadata that can be acessed directly from images.
* Added an article on object detection to the user guide.
* Added the `limbo-compress` command for optimizing data acess.
* General code and documentation cleanup.

* Completed one million synthetic images!
* limbo-stats generates additional statistics, including license and copyright information.
* Categories can contain slashes.
* Added the limbo-lint command for cleaning datasets.
* Switched to pyproject.toml and flit for packaging.
* Greatly improved install documentation.

* Initial release.