Improve rate limit handling

I changed the error handling for queues, but not sure what to do with the Batch API method. Error handling used to be that errors are logged but end users aren't notified. Do you still think this is the way to go?

Hmmm we've never used it at scale in bulk so never had the need yet, but really good feature! Seems like better to show the errors to the user then if run via batch. A rate limit error would hint to them to manually try again later mimicking the automated delay of queue worker.

Looking at your MR at a glance the delay might be a month? If so I wonder if you'd get support requests from your clients asking why it's not generated yet. Would it make sense to add a messenger status to node edit like 'PDF generation is queued for this content item to be processed after (date time)'?