Initial UFSDataset

[timothyas]

This adds the module src/UFS2ARCO/ufsdataset.py with the class FV3Dataset that can be used to read replay data from s3. Some other contents in this PR

• docs/read_and_store_replay.ipynb: gives a quick example usage of this functionality, and a brief look at xarray/dask/zarr interplay
• scripts/read_from_s3.py: gives another example, this one pulling multiple forecast cycles into a single zarr store
• scripts/config-replay.yaml: the yaml file used in that example
• environment.yaml: I suggest we use this environment file since it is slightly more minimal, and potentially easier to maintain than the one in src/ . With the addition nbsphinx package, we can add notebooks like the one added above to documentation.

The notebook I added only takes a shallow dive into how xarray/dask/zarr all work together. I'm happy to add more as necessary, please let me know if there's stuff that's not clear.

[timothyas]

Also, I'm not sure how you guys feel about merging versus squash-merging, but I suggest we (eventually) squash-merge this PR. I'm a fan of micro-commits as I work :) although this can muddy up the project history