Retry Extraction

[norberttech]

During implementation I realized that RetryExtractor is not the best idea, since we have no control over single batch, only whole extraction.
This means that if we need extract 10 rows one by one and row 8 extraction fail, RetryExtractor would retry the entire extraction duplicating rows 1-8.

Instead we need to create a simple retry mechanism (based on the building blocks from this task) and pass it as a dependency to extractors, like this:

(new Extractor())->withRetry(retry_operation(...))

Originally posted by @norberttech in #1842