Thank you :-)

I'm wondering what's wrong here and if it's possible to mitigate that.

This is an interesting case.  Incidentally I've been working on a large rewrite which spans also code responsible for this failure.  Fixing this case was easy now, after that refactor.  I hope to deliver it to releases soon, probably 5.x as well - but if you're starting with Knot Resolver, I strongly recommend skipping 5.x versions.

--Vladimir