Make sure all datasets used in tutorials are easily acessible

[Marius1311]

Description of feature

Currently, datasets used in tutorials need to be manually obtained by the user, which creates a hurdle to running the tutorials and experimenting with the HNOCA tools. Ideally, all datasets used in tutorials would be on FigShare or Zenodo as they are, and accessible via the AnnData interface with

adata = sc.read(
    adata_path,
    backup_url=DOWNLOAD_URL,
)

such that the first time this code is executed, the dataset is automatically downloaded, and subsequent function calls will import locally from disk.

[Marius1311]

The same goes for auxilary files, like the marker gene yaml file used in the annotation tutorial. As that's a small file, it could even be part of this repo.