[FR]: Enable better backoff/retry handling outside of robust_getitem

[adkinsjd]

Feature Summary

Xee uses xarray's robust_getitem to handle errors and retry when fetching the data itself from EE, however other calls to the EE API, such as for initialization and get_info do not catch EE API errors and retry them, making the running of large/long download jobs from EE difficult.

This especially arises with 429 rate limit errors. Ideally all API calls would back off to an acceptable rate, not just the pixel data fetcher.

Use Cases

Downloading large datasets or running large jobs on EE data that run up against EE rate limits without hand-tuning the size of the Xarray workflow to match the underlying EE maximum request rate.