A Fujitsu RAF file from a Super-CCD sensor camera contains two images. One has the data from the larger more sensitive "S" sensels, the other for the smaller, less sensitive "R" sensels. These are supposed to be HDR combined while processing the file. Please refer to the discussion at https://translate.google.ie/translate?sl=auto&tl=en&u=http%3A%2F%2Fwww.g... for an overview of this.
As far as I can determine libraw doesn't merge the data from the 2nd image at all, but simply returns (by default) the image data for the "S" sensels. AFAICT it is possible to retrieve the "R" image by specifying imgdata.params.shot_select =1
Do you think there's any chance that libraw retrieve both images and merge them to present the combined data?