This MR adds a new algorithm which is an improved version of the
Timepix3MaskCreator as it works for arbitrary detectors and uses the
mask_file provided along with the detector in the detector configuration.
Currently it only contains the same mean frequency cut-off approach as the other mask creator, which is okay if the sensor is evenly illuminated. With inhomogeneous beam spot, this will lead to issues and we mght want to consider implementing the local density kernel approach from proteus.
Maybe we could keep both and allow selection of the method.