I agree that even if release of data were not ordered, there would still be plenty to go at. PACE does not include the severely affected and apparently shows CBT/GET to be ineffective in others. There are interesting calculations somewhere, I think by by Anciendaze, which showed how unrepresentative the sample included in the trial was.
If those figures are combined with risk of harm for some sufferers from exercise, it should be possible to construct a case to show that it would be unethical to impose CBT/GET without first undertaking studies to distinguish between those who might benefit and those who would be harmed. It may be the insurance companies that will have the last, as well as the first word, on this matter.