Singlecellexperiment

- Rename `reduced_dims` to `reduced_dimensions`. Constructor accepts both these arguments for backwards compatibility.
- Additionally, SCE's constructor takes kwargs for future proofing any additional attributes in the upstream classes.

- Support reading outputs from CellRanger version 3 and later in `read_tenx_mtx`.
- Handle both `genes.csv` and `features.csv` for gene annotations.

- Check if the column names of the alternative experiments match with the column names of the main experiment. This is the equivalent to the `withDimnames` parameter in the R implementation.
- On **getters** of alternative experiments, if `with_dim_names` is True, column names of the alternative experiment are **replaced** with the
column names of the main experiment.
- On **setters** of alternative experiments, if `with_dim_names` is True, column names of the alternative experiment are **checked** with the
column names of the main experiment and an Exception is raised if they do not match.

- Add wrapper class methods to combine experiments by rows or columns.
- Expand function names for readability, still backwards compatible with the older function and method names.
- Add getters and setters to replace a specific alternative experiment or reduced dimension.
- Fixed an issue with numpy arrays as slice arguments. Code now uses Biocutils's subset functions to perform these operations.

- chore: Remove Python 3.8 (EOL)
- precommit: Replace docformatter with ruff's formatter

- Fix package version issues to support Python<=3.9. Mostly related to how anndata dependencies are versioned in the MuData package discussed [here](https://github.com/scverse/mudata/issues/82).
- The package now enforces the versions of mudata, anndata and numpy that are compatible with each other.