Enable S3 Bucket Transfers

[DavidHuber-NOAA]

Description As workflows migrate to cloud platforms, it would be useful to add an interface to write to and read from S3 data buckets as a convenient way to keep archived data in the cloud. Extending the capability onto the fsspec library would be an ideal solution.

Requirements

Data files can be transferred to and from S3 data buckets with using a similar interface to similar, existing functionality (e.g. hpss.py, hsi.py, and file_utils.py).

Acceptance Criteria (Definition of Done)

• [ ] A class is written with read, write, and verify methods
• [ ] Tests are added for each new method

[edwardhartnett]

Note that netCDF can transparently write/read to the cloud. In other words, instead of copying files to the cloud, you could add NC_ZARR to the create mode flags of your netCDF nf90_create() call, and create the S3 file with existing netCDF code.