Fix hanging processes when warc download fails

[rfbr]

Previously when a RequestException occurred while getting a WARC url, the sub-process running the document_generator function would crash without yielding any result.
The following logic

while True:
    ...
    if progress_bar.n == len(warc_paths):
        break

would thus never reach the expected count, preventing the loop from breaking, causing the main process to hang indefinitely.

This commit fixes the issue by:

• adding a WarcResults dataclass to track success status
• adding the get_response call in a try-except to catch RequestException, yielding a WarcResults with success=False in case of a RequestException
  It also improve the error reporting with some statistics about failed WARC and records.

[gabrieldemarmiesse]

Can this be an attribute computed dynamically to avoid asking the caller to keep track of it?
Something like

@property
def total_records():
    return self.processed_records + self.failed_records

[rfbr]

Good idea! Done.

[gabrieldemarmiesse]

Here we never print the exception, it can be a pain to debug then. Can we put it in WarcResults? If it's not pickable we can always store the error message there.

[rfbr]

As it was mainly RequestException or IncompleteRead errors I thought it wasn't really needed. Nevertheless it makes more sense to log the error. Done!