if you split and mirror as before but now move in two "beam splitters" and 4 detectors to determine if one is always reflected and one is always transmitted if the path lengths are kept the same.
Mine is defferent as it takes out the recombination possibility by cutting the pathe short -- take the original diagram and put the beam splitters at half the distance from the mirrors and move the detectors up accordingly. you end up with four possible end paths and four detectors to test them.
that doesn't give the same control on the experiment. I want to see it they always go to the same side even if they do not have the {combination\interference} possibility.