I tried debug-on-entry before calling dired-jump and didn't see any errors. I am not sure if I know what I am looking at when I drill down to the active variables. As I mentioned, it is not completely frozen, it is just taking a very long time when there is a large pdf in the buffer, so it might not throw an error.
But I did do another test which seems to rule out helm in itself as the problem. In an emacs with no packages except pdf-tools and dired+ the same problem occurs in ido-mode, but NOT when simply using the default basic completion style. Ido-mode offers completion candidates, but just as with helm, when you try to go to a dired buffer, emacs hangs. With no dired+, the same operation in ido-mode works smoothly. So it seems the issue is being caused by dired+ in combination with not just helm, but ido completion. I could try other completion packages, I suppose, to see if those also have the same problem.
Sorry, last message here. I will follow up with diredp-send-bug-report. I will try your last suggestion.
Thanks!