Pytximport

API
- Added a new count transformation for differential transcript usage analysis: `dtu_scaled_tpm` (`dtuScaledTPM` in R-version of `tximport`)
- Added a utility function for renaming transcript-level exports with the transcript name: `utils.replace_transcript_ids_with_names`.
- When using biotype filtering, the TPM values will now be rescaled to match the sum of the TPM values before filtering.
- Apply biotype filtering, ignore transcript version and ignore after bar for transcript-level exports as well.
- Bug fix: Saving to files when exporting transcript-level counts works now.

Other changes
- Moved transcript version removal into the utility function `remove_transcript_version` and biotype filtering into the utility function `filter_by_biotype`.
- Always run unittests locally for each commit with pre-commit.

API
- Change the default `output_type` to "anndata".
- When a biotype filter is used, the abundance of the remaining transcripts after filtering will now be rescaled to match the total abundance pre-filtering for each sample.
- Bug fix: Now returns AnnData objects when returning transcript-level data if `output_type` is "anndata".

Other changes
- Run CI on `dev` branch.

API
- Automatically change `output_type` to `anndata` if `output_format` is `h5ad` and warn the user if a different `output_type` was provided.

Other changes
- Add a unit test that tests for correctness against the R version of `tximport` (version 1.32.0).
- Release versions via GitHub actions.