Matches the four sectors of the detector to a common level.

This routine uses the average of the top and bottom reference pixels to establish the global supermean level of the ramp.

Then the reference pixels (top&bottom) of each sector&frame are tuned to match the global supermean.

This routine performs an operation similar to the zero (first) read subtraction, with some advantage a) it preserves the DC level of each frame, useful for linearity and saturation check. b) facilitates the noise analysis b) since images are not differential, the anomalies (e.g. cosmic rays) of the first frame do not disappear but propagate through the ramp. There is also a disadvantage for quick look, as the images maintain the fixed (bias) pattern that gets removed by the zero image difference. The bias pattern, on the other hand, will be more correctly derived by the line fitting procedure.