Thank you both again for trying to help!
@contradict guess we should have kept on googling a bit more. That should solve our second problem. Also thank you for refactoring the code (I just learned new Julia syntax 
). Also, you’re right about the Float64 conversions, your version handles that much more elegantly. Actually, I added those only after upgrading to 1.4.1 because of some other type-issues.
Coming to the most important part of your reply, I can confirm that this solution also works on my machine! I think I’ve made this clear by now, but this helps A LOT 
:
I went through your version and everything seems to be implemented correctly from an algorithmic perspective as well. Still not sure, which of your alterations actually did the trick, but I am fine with that for now. Thank you again.
Do you think this might still be worth filing a report for?