In terms of UI, definitely still some room for improvement there. In our changelog we do list it as a beta feature as well, so it may not be full ironed out.
Regarding how much it does at one time, it defaults to 10 at a time, and grabs only posts that do not have the meta data values we include to help determine status. It also run on a WordPress cron schedule of every 10 minutes. Done in lower volume at a time to help accommodate for various hosting environments and try to avoid any timeouts etc.
Correct on the first 800 character limit. We’re considering bumping that a lot, or adjusting how we grep content as a whole, with Gutenberg now out in the wild, and character usage that the user isn’t going to see from there for its own parsing. 800 characters would get gobbled up fast and cut things short.