Hi rtemen!
Tried to analyze your issue.
Cant reproduce exactly your workflow as during calibration you used
Master_Bias, Master_Dark and Master_Flat2 (!!!).
None of them are accessible for me on the dropbox.
However,
Find attached a screencopy to see what i tried with your submitted data.
Find in the first upper row at left a diagonal profile cut through
your single flat, right beside the profile through my corrected Flat version,
and at most right a profile through your so called Masterflat which is more or less unusable
because of the very dark corners.
In the second row at left is the calibration result using your single flat and single light, you see because
of the flat beeing to dark you get bright corners.
right beside in the middle the result using the single "corrected flat" which gives a well calibrated light (c_3).
What did i do? I added a pedestal of 0.08 to your single flat and saved it as f2.
(Above is the corrected flat profile cut and the PicelMath expression to correct your original flat)
In the second row most right is the calibration result when using your "MasterFlat". Clearly it has
to much bright corners as the corners in the Masterflat are to much darker than the centre.
In the 3rd row, at left your debayered result, in the middle my debayered result.
In the 4th row bottom left your original light, in the middle the histogram of the corrected flat f2.
Here the link to the screencopy
http://www.werbeagentur.org/oldwexi/flatcorr.jpgGerald
P.S.: For me it looks like you had different bright single flats which lead to the funny masterFlat.