Going to stick my neck out here Peter - please forgive the impertinence
.
The results would be consistent if the blue2 and green2 filters had been inadvertently switched - could this be possible?
If possible, you would have:
RedDS =866, Red2 = 745, Ha12 = 717, Ha6 = 664
GreenDS = 0, Green2 = -3
BlueDS = -1097, Blue2 = -1139
This could suggest that the filters could possibly be pretty close to parfocal - as you note, it is pretty hard to make them with offsets anything like you measured.
If the two filters were actually swapped, the offset due to OTA CA is maybe a bit more than expected, but not ridiculously so.
Cheers Ray