Updating n-arm DetectorMaps leads to silent problems

For this reduction, using

reduceExposure.py /work/drp --calib /work/drp/CALIB --rerun rhl/eng-2023-07 -c isr.doFlat=False isr.doFlatNir=False doAdjustDetectorMap=True doMeasureLines=False --id visit=97083 arm=b^r^n spectrograph=1^3

all 6 arms "succeeded", and br look fine, but here is a part of the 1200 n-band spectra image.

Among other weirdnesses (the lines should all be straight and vertical), It looks as if something happens at the centre of the n1 chip. The HgCdTe detectors are monolithic, unlike the CCDs, so all the pixels should be solved together. I'm not sure that that's what's going on, though. Even in the optical the only thing that the two CCDs can do is differ by an affine transformation (or actually probably only a translation/rotation/scale).