I'm saying the details of how they generate that figure doesn't matter, what matters is its information content, which I can get in much easier ways. Let me ask if you agree that the "average trajectories" that they plot are indeed exactly the same as we would get via my method #2 above-- running one photon at a time through exactly their configuration, and just putting the wall at different distances, and collect the aggregate detections. Then build up a concept of the aggregate photon flux by taking those measurements, normalizing the total detection numbers to be a constant total for every wall distance used (zero divergence), and then drawing the "field line density" for that divergenceless detection field? That's exactly how we would generate a concept of "aggregate photon flux" in this very two-slit experiment, in a completely classical limit of many iterations of slightly different experimental setups (the distance to the wall being the sole variable).
If we can agree that I can get the exact same figure my way, with no subtle "weak measurements", then the question to ask is: what additional information are they extracting with their clever measurements if they end up with the exact same figure I get?
Note that it makes no difference how clever their measurements are-- if they can tell which slit the photon went through, they won't get that photon to participate in an interference pattern anywhere. That is all the CI needs to hold.