A study found fraud rampant in retracted journal articles -- and publishers less than transparent about why articles were pulled