Lambda the Ultimate

The lambda calculus is not upward confluent, counterexamples being known for a long time. Now, what about the interaction calculus? Specifically, I am looking for configurations c₁ and c₂ that have the same normal form with no such c that c →* c₁ and c →* c₂.

Update: a necessary and sufficient condition for strong upward confluence discussed in arXiv:1806.07275v3 which also shows that the condition is not necessary for upward confluence by showing upward confluence for the interaction system of the linear lambda calculus.