tl;dr: for 3x3 images, please add an X offset of 2 pixels in order to get the images automatically processed by the reduction pipeline.
As you might have seen, we reactivated the pipeline and also started taking FLATs again, so that you can download fully reduced images.
For this to work, your images must have exactly the same dimensions as the calibration frames. This works fine in 1x1 and 2x2 binning. But in 3x3, unfortunately there is a little problem: The first 50 (unbinned!) pixels on the CCD are covered, so that they effectively serve as DARKs. Now we don't want a binned pixel to cross this border. This means that for 3x3 binning, we take the calibration frames with an X offset of 2 pixels, resulting in 48/3=16 binned pixels in the covered area.
Long story short: for 3x3 binning, your CCD settings should look like this:
Note: your reduced images will not undergo any dark/bias or flat correction, if the image dimensions do not match!