I suspect that the CFS-only subset was too small relative to the size of the other sample sets to safely draw the conclusion that CFS-only patients don't have recovery problems while CFS+FM patients do. I'd want to see some statistics... maybe some calculations of error...?
And, as always, the fuzziness of diagnosis raises questions about the accuracy of subset-specific conclusions. For example, how did they decide whether patients have comorbid FM? There's no objective measure for FM or CFS. If they included all "CFS" patients who had muscle pain in the CFS+FM group and only non-pain "CFS" patients in the CFS-only group, then it's possible the 18 CFS-only patients don't even have ME/CFS as we know it. Maybe they are depression or hypothyroid or goodness-knows-what-fatiguing-illness patients.
I've had some ME/CFS specialists say I have comorbid FM, and others say I don't. My GP diagnosed me with FM, but I honestly don't think he knew that there's a difference between FM and ME/CFS.
Frankly, I don't believe I have comorbid FM at least in part because my pain appears to be of an entirely different nature than that described by FM patients, but who knows for sure?
IMO, the diagnoses are far, far too unclear to be drawing any conclusions about CFS-only vs CFS+FM.