Previously FLP was uploading files that users can read with a web browser -- HTML, PDF, and also XML with metadata. I could and did link directly to HTML and PDFs, including circulating these materials with coauthors and research assistants and members of the press.
If FLP begins uploading only huge tarballs, and not the individual constituent files, I won't be able to do any of that.
Perhaps the "only" is telling here. Were they previously also uploading the tarballs? No wholesale user would want to scrape the thousands of extra pages of HTML to download the content. So if they weren't already uploading the tarballs, this is actually a beneficial change.