gmultiset.gmultiset_eq_dec applied too eagerly?
I don't have time to debug this in more detail so I am just going to dump a representative trace here. Without exception, all my diverging type class searches (of which I have a few per hour) eventually look like this one: trace.txt
(Just in case somebody actually looks at it: the piece of code you can see in line 1 doesn't actually make any sense, I think.)