The fetchlimit system test failure reported in #69 (closed) happens very rarely and we currently have no test artifacts to look at. This MR improves failure logging in the relevant subtest in order to bring us closer to an actual fix the next time the issue manifests itself.