Muvi

[Hotfix NaNs in prior masks.](https://github.com/MLO-lab/MuVI/commit/1957ed4690667fdf82d2b63c053576dd71b825f0)

Summary

This release includes enhancements, bug fixes, and new features to improve functionality and usability. Key updates involve fixing renaming issues after reordering, updating plotting defaults, adding new functionalities, and addressing bugs.

New Features

- **`muvi.pl.missingness_overview`**: Added a function to provide an overview of missing data.

Changes and Improvements

- **Allow Factor Sorting by R2 and Renaming by Enrichment Significance**: Enhanced factor sorting and renaming capabilities.
- **Set Default Renaming to False**: Adjusted default to set renaming to false for better control.
- **Redefine `n_factors` as Number of Uninformed Factors**: Updated definition for accuracy.

Bug Fixes

- **Fix Renaming After Reordering**: Resolved issues with renaming operations post-reordering.
- **Fix Variance Explained Grouped**: Fixed unwanted reordering with grouped variance explanation calculations.

Skip normalization for Bernoulli likelihood.

Release Summary

New Features and Enhancements
- **Data Handling**:
- **Deduplicate Indices**: Remove duplicate indices to ensure data integrity.
- **Data Merging and Normalization**: Implement methods for merging (union/intersection) and normalizing data.
- **Flexibility in Masks and Samples**: Improve handling of missing/extra features in prior masks and allow redundant samples in covariates.
- **Integration with `adata` and `mdata`**: Update `from_adata` and `from_mdata` methods, including support for `obs_key` to load observations from `obsm` if required.

- **Plotting and Visualization**:
- **Group Plots**: Introduce group plots to facilitate visual analysis of grouped data.

- **Model Loading**:
- **CPUUnpickler**: Add `CPUUnpickler` to load models trained on GPU, enabling model compatibility across different hardware setups.

- **Feature Set Enhancements**:
- **Posterior Feature Sets**: Allow computing and exporting posterior feature sets using [kneed](https://github.com/arvkevi/kneed).

- **Miscellaneous**:
- **TensorDict**: Add `tensordict` support for more efficient mini-batching.
- **ASCII Representation**: Add ASCII representations of data for better readability and debugging.