When URL is a retweet or quoted tweet, use the originating tweet's timestamp for first_occurrence

The ultimate goal here is to try to determine how old a particular news item is. Similarly, it would be cool if we could extract the publication date of articles from their pages, but this is almost certainly more involved. Looks like the [newspaper](https://github.com/codelucas/newspaper) library could be a good option with the `.publish_date` attribute. Actually, yeah, the more I look at it, the more interested I am in the newspaper library. Make that a separate issue.