When running at scale, sleep in code can have dramatic affect.

[mike-gangl]

https://github.com/SWOT-Confluence/ssc_input/blob/e9930c51e78673d2b09bd569674d6ea30e8c5d78/ssc_input.py#L133C9-L137C39

These sleep commands, which will almost certainly happen as we are hitting CMR with many queries, cause constant delays (up to two minutes). For reference, we have ~200k reaches to process, so at 1% of those hitting a CMR Error, we might be looking at (assuming average) a 2000 minutes of delay in SSC_input processing.

We can respect the 'retry-after' header in the CMR Queries:
https://cmr.earthdata.nasa.gov/search/site/docs/search/api.html#request-moderation