@christoph-v-kontron - I thought of it couple of time and I could not decide what is the right approach. It is a very particular case it happens in your scenario with a 200ok suspended.
Somehow make sense what the patch tries to do, but the side effects are not completely asserted. I would say that we should add a modparam to control this behaviour at the moment, with a default value making it to behave as the patch proposed. But still to have the option to turn it of and behave as before the patch.
So I am going to merge the PR, then I will add a new modparam to be able to control its behaviour.