I also dislike #4 from that list for the reasons stated there. For example, if you say WHEN MATCHED AND x.some_boolean and then WHEN MATCHED, you expect that every tuple that hits the latter clause will have that Boolean as false or null, but #4 makes that not true.
I think the best options are #2 and #5 -- #2 because it's simple, and #5 because it's (maybe) more intuitive, albeit at the risk of livelock.
As you said, #5 seems the best and that's what the patch does. But ISTM that the options you listed are not really the concerning points. As the patch stands today, we evaluate WHEN AND conditions separately, outside the EPQ. The problem arises when the join qual returns a different result with the updated tuple. I listed down those cases in my earlier email in the day. To me (and I assume to Peter and Simon too), those are the more interesting cases.