// TODO: can use more cores than coresToUse()
static IBWIntegralImage stripedBWIntegralImage(ThreadPoolExecutor executor default null, int stripes, BufferedImage img) {
  int minPixelsToSplit = 16384;
  int w = img.getWidth(), h = img.getHeight();
  if (stripes <= 1 || w*h <= minPixelsToSplit)
    ret BWIntegralImage(img);
  else
    ret BWIntegralImage_twoParts(executor, img, (img2, idx) ->
      stripedBWIntegralImage(executor, (stripes+idx) >> 1, img2));
}