adding an extra step to allow segmenting the double flatfield images

I have added a new step to create a binary map and isolate strong defect visible on the flatfield image. This map is applied to the Double flatfield images, ensuring only parts with defects are corrected (limiting over correction where not needed). The thresholding is a new parsed argument, the value to parse represent the upper percentile of the histogram used to segment defects. At the moment, if we request 0.01, it's 1% percentile, actually segmenting the images for all values above 99% percentile of the histogram. Maybe it would be better to add a security to ensure no value > 1 are parsed (we could add a test, if > 1, then divide by 100). Also, if the fact that 0.01 means "keeping only pixels above 99%" is confusing, we can change the logic. I also change the vertical median when multi median angle is requested: as this is quite computing heavy, I force the vertical median to be on a fixed window rather than a sliding window.