Feature/anndata io

[minnerbe]

This is an attempt to include AnnData support:

• Datasets can be in either the pre-existing N5-based layout, or an AnnData-conforming layout.
• For both layouts, all backends supported by the N5 API (FileSystem, HDF5, Zarr) can be used, although some may not work properly until N5 can write String-datasets; see also this PR.
• Containerization of datasets is now optional, but alignment still needs a container due to the inherent multi-dataset structure of the task.

Major changes in the code:

• Tests were added for all parts of the code that were changed.
• The io module underwent a major re-design, implementing the features described above.
• The documentation was adapted to the new situation.