Ok, here's some more detail on the problem and a work-around:
I noticed that the MasterBias and MasterDark were listed in the script file boxes as "Binning 1" when they, like all the other frames were actually 2x2 binned. These two masters had been made by my former calibration method in PI that followed precisely Vicent's Calibration tutorial, as I had captured in a series of calibration process icons. So those masters had previously been created by ImageIntegration, but for some reason the preprocessing script was calling them "Binning 1" instead of "Binning 2" as it had correctly classified all the other input files. Not sure this was the same or yet a different problem than whatever caused it to look for 2-second dark frames to calibrate the flats above, but when I removed those two masters as inputs and replaced them with the original bias and darks frames used to make those masters (unchecking masters), the preprocessing script hummed away and worked just fine.
It would seem there are a few loose ends here...
Thanks for taking a look. And thanks for a great script that saves a lot of time.
Clear skies,
-Jeff