Paul,
If by "happened" we understand what was observe, what has observable consequences, then we agree 100% that what "what happened has already happened". The only difference may be in the order of the events. In classical mechanics, time is linear, in quantum mechanics, in some cases, is not, as I will explain. Time is a parameter, very similar to how space coordinates are. In classical mechanics, the events happen as time goes. In other words, all the events labeled with time smaller than t already happened at time t. Time is linear. In quantum mechanics, the things are really different, in the sense that the order in which the events happen may be different for that of time labels. In the case of the Mach-Zehnder interferometer, the photon leaves the first beam splitter, and arrives at the place where the second beam splitter may or may not be. The interaction between the photon and the second beam splitter takes place or not (if the beam splitter is not there). This is the happening at the second beam splitter. Then, and only then, the happening at the first beam splitter takes place. The happening at the first beam splitter, so to speak, stays suspended, until the happening at the second beam splitter gives enough information to the system, so that the photon will know whether to go one way or both ways. So, I agree that "what happened has already happened", except that, for quantum phenomena like the one discussed, the order is not the linear order of events. If we want, we can say that the order is linear, and the happening at the first beam splitter took place before the happening at the second one, if we assume that the photon can guess the future. Even in this case, the causal order is reversed, because the photon's "choice" is decided by a future event which it guessed.
Best regards,
Cristi