Add code to generate Zarr dataset

[jeromekelleher]

Once #270 is done, we can add some more steps to the pipeline to create the Zarr dataset.

This will

1. Download the metadata and description json from Figshare
2. call ``sc2ts import-alignments dataset.zarr [alignments/*.gz] Add support for reading raw MAFFT alignments sc2ts#460
3. call sc2ts import-metadata dataset.zarr metadata.tsv.gz --descriptions=field_descriptions.json (requires Add support for "description" for metadata fields sc2ts#459). We may want to do this in Python within snakemake so that we can call the massage_viridian_metadata function.
4. (Optional, probably doesn't make much difference). Call Dataset.reorder by the Date_tree field.
5. Call create_zip to create the final version.

With this, we should have a 100% reproducible and automated pipeline for generating the final mafft-aligned Zarr dataset from the primary sources.