These scripts you're talking about, could they be automatically generated for all master disks or did it involve some manual work? If the last is true, then that sounds like IPF generation where you not just copy blindly, but try to describe the format on a higher level to be able to verify it properly.
Isn't flux level disk copying similar to creating .TAP files from C64 tapes? Here you also look at zero-crossings in the source signal and create timing codes from them. If you rip the same tape in two different tape decks (or the same for that matter), the output will not be 100% identical, although they might both load just fine. Then there's tools like TapClean which can clean it up and in many cases make two rips be identical, but this requires knowledge of the loader being used. It will not be possible to clean it up completely if you're blind.
Unless there's an inherent hardware quantization in the timing of flux transitions (is it being latched with a clock in the floppy drive?), then it should be considered as an analog signal.