Guidance on Zarr benchmarking

[abarciauskas-bgse]

I've started to work on guidance for Zarr chunk size and shape. A future version should link to or include guidance on how to benchmark different zarr chunk sizes by making it easy to generate zarr stores with different chunk shapes and then execute tests against a configurable library.

Some prior and current art to review:

• Benchmarking Zarr and Parquet Data Retrieval using the National Water Model (NWM) in a Cloud-native environment
• rabernat/zarr_hdf_benchmarks
• Benchmark Zarr Implementations