Thank you for uploading the frames. These are my findings:
1. I confirm that no bayer offset keywords (XBAYROFF/YBAROFF or BAYOFFX/BAYOFFY) and no ROWORDER keyyword are present in the FITS header, but this is NOT a problem. Since no ROWORDER keyword is detected, the setting of option 'USE ROWORDER keywords' in FITS Format Preferences does not take any effect - contrary to your statement in post #4.
2. With setting 'Coordinate origin' to 'Upper left corner (top-down)' in FITS Format Preferences, a true sided image is obtained for the M81 frame; the astrometric solution says:
Rotation ................. -91.197 deg
North is at the left, west is at the top, the image is NOT flipped. This result is to be expected with a refractor (number of mirrors: 0) when the correct vertical orientation in FITS files was set.
3. With setting 'Coordinate origin' to 'Upper left corner (top-down)' in FITS Format Preferences, a correct Debayer result is obtained when option 'Bayer/mosaic pattern' is set to 'Auto'.
So in my view, everything is fine.
Recommendation for correct settings:
In FITS Format Preferences 'Coordinate origin' shall be set to 'Upper left corner (top-down)', and option 'USE ROWORDER keywords' should be enabled (these are the default settings).
In the WBPP script, global option 'FITS orientation' shall be set to 'Global Pref' (the default setting).
I want to stress that if you formerly used a different setting for 'Coordinate origin' in FITS Format Preferences or for 'FITS orientation' in the WBPP script, you must not use old master calibration files. You can use old calibration subframes, but new master calibration files have to be generated therefrom.
Bernd