Hi - I see that Paul has already added a lot of thorough input, so I will only add this big top-level question:

Why does this PR implement a retry mechanism manually (using hardcoded 1000ms and a handwritten backoff strategy) rather than rely on the mechanism provided by the API itself? See doc at https://docs.github.com/en/rest/using-the-rest-api/rate-limits-for-the-rest-api?apiVersion=2022-11-28. Especially the section "Best practices for using the REST API > Handle rate limit errors approriately" has a lot of guidance that this PR is not implementing, but could be. (The exponential retry is only recommended as a last resort, rather than the first cause of action!)

@RudolfSchreier Do you have some example code which follows the github blessed retry mechanism?

@RudolfSchreier I've changed the retry mechanism to follow gh best practice using incremental backoff as last resort.

@RudolfSchreier I've changed the retry mechanism to follow gh best practice using incremental backoff as last resort.

I've synced on this with @roosnic1 yesterday and am also a bit swamped with work - please feel free to move forward on this without my required approval :)