New features
- Introduced CLI parameter [--X-approximate-distribution](https://github.com/chanzuckerberg/cellxgene/issues/2282) for users to specify the distribution of their data, and override the default heuristic determination
- For categories with the number of labels over the number set by --max-category-items (default 1000), we now include a label that aggregates all labels over the limit instead of truncating the list of labels
Breaking changes
- Removed support for experimental features (--experimental-annotations-ontology, --experimental-annotations-ontology-obo and --experimental-enable-reembedding)
Bug fixes
- Fixed a [bug](https://github.com/chanzuckerberg/cellxgene/issues/2211) in differential expression calculation by changing the t-test calculation to consistently assume normally distributed data, and to log data when data is not normally distributed based on a heuristic
- Fixed a [bug](https://github.com/chanzuckerberg/cellxgene/issues/2225) in --max-category-items