Potential Improvements to CC download & Extract Stage

Is your feature request related to a problem? Please describe.
The CC download and extract can have better user experience (and potentially be made faster). Here are a list of things that can be explored:

• Add retry logic to download #1225
• Retry logic for url generation/saving CC url's? Ran into a timeout from CC trying to get URL's. If they've already been pulled they can potentially be saved someplace to retrieve later?
• Support saving CC downloads to cloud sources. Each snapshot can be 80-100TiB in size and is difficult to store locally #779.
• Explore row level parallelism for extraction. Each core/CPU extracts all rows for a warc file serially. This needs to be explored further but there are often some rows that may take a much longer time to extract than others (deeply nested html files). Extracting at a row level might help get the best performance from extract which is relatively slow.
• Limit Download concurrency for Ray-Data

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

Start by locating the Common Crawl download and extraction pipeline, including the Ray-Data entry points mentioned in the issue. Review the existing download, URL-generation, cloud-storage, extraction, and concurrency behavior before narrowing this broad list to one scoped improvement. Done should mean that the selected behavior is implemented and its performance or retry/concurrency outcome is verified.