The problem with this argument is that it assumes that (a) CFS is a well-defined single condition, and (b) the detection of XMRV by those researchers who are finding it is 100% accurate with no false negatives.
Since "CFS" is clearly not well-defined, and there are almost certainly going to be misdiagnoses in the population diagnosed "CFS", any arguments based on such percentages are based on a false assumption. It just isn't useful to make those sort of arguments about percentages found in CFS studies based on the assumption that "CFS" is a single entity. That kind of spurious argument has been used greatly to suppress research into ME/CFS. It's such disingenuous behaviour it's hard to credit that it isn't done deliberately, and yet when one talks to doctors and supposedly intelligence people about CFS, one discovers that people really are that stupid.
More importantly in this case, though, is the fact that XMRV detection is in its infancy. To achieve 75% detection rates is extraordinary enough at this stage of understanding; there's no reason to believe this figure won't increase. Indeed an as-yet-unpublished paper by the WPI is said to find 95% infection rate in the patients they studied, probably reflecting their expertise in distinguishing "true" (XMRV) ME/CFS as much as anything else.
More importantly still, XMRV is believed to be a "stealth" retrovirus, lying dormant in infected tissue until triggered by factors like levels of stress, hormones and inflammation. So XMRV can easily be lying dormant in individuals that test negative, and the controls who test negative could well develop ME/CFS in the future.
Having said all that, I think one thing we can say is that based on the early data, it would appear that only roughly 10% of those infected with XMRV are currently diagnosed with ME/CFS (assuming 7% general population with XMRV infection and 0.7% with ME/CFS). So this does indeed suggest to me that - as the WPI and Dr Coffin have said since day one - there very probably are co-factors influencing ME/CFS development, and as just one simple example, it could be that only about 10% of the population have the relevant (innate or acquired) deficiencies in RNAsel or APOBEC3 pathways which mean they are unable to 'reprogram' XMRV infection.
So in summary: yes, it does seem to me likely that there are cofactors, however it also seems to me that in the terms we normally use for disease progression, XMRV is more likely than not "the" cause, or the "sole" cause. If a person were poisoned by an overdose of radiation, we would not emphasise the fact that their symptoms were not solely caused by the radiation, but were also "caused" by the lack of genetic factors that might have protected them better. Such factors will always apply to any pathology, but that doesn't change the fact that in this example, the radiation was the "cause". (It's not a great analogy but I hope you get the point.)
At this stage we just can't assume very much. To achieve rates of 80%+, XMRV seems to be fundamentally significant, but beyond that, the detail remains to be uncovered. There may or may not be cofactors; there will definitely be other factors that are relevant but whether those will constitute a "cause" is another matter.
The most significant question for the general population regards the 3-8% rate in the control groups. If XMRV has been around for 100,000 years, then that does imply that 90% of people with XMRV don't get ME/CFS - they might all be on their way to prostate cancer, MS, alzheimer's etc though. On the other hand, if the XMRV connection to 2 strains of lab mice is the result of a laboratory origin to XMRV, then we are looking at a new epidemic within the last 100 years. If that is the case, then the 4% rate would represent the "next wave" of the pandemic (those people or their descendants may be on their way to neuro-immune disease), and would suggest that rates of ME/CFS will rise in the future from 0.7% to 4%. That would be a consistent epidemiological picture in the 'modern epidemic' scenario.
The idea that there are cofactors with XMRV that determine ME/CFS progression is not a problem. But the people who try to emphasise that XMRV may be a "coinfection" or "passenger" are rather more problematic. Logically, their position is consistent in that it is a theoretical possibility, but the details of retroviral activity and of XMRV specifically mean that this would be an extraordinary coincidence if so. The people who are arguing strongly about "passengers" and "coinfections" are mostly doing so based on their existing assumptions and interests about ME/CFS.
Bear in mind that the psych lobby want to proceed from the assumption that ME/CFS is a primarily psychological condition, and to argue that immune vulnerabilities can be a consequence of personality traits and negative ways of thinking, and to therefore argue that any infections, viruses, or immune vulnerabilities are ultimately a consequence of psychological factors. In that context, with the psychological model being the predominant model for many professionals, the "coinfection" assumption will inevitably be the first argument that will occur, and the one that many people want to be the case.
But that just explains why "coinfection" and "passenger viruses" have received so much more attention than logic deserves; at the moment the most that can reasonably said is that those ideas are still "theoretical possibilities", albeit rather implausble ones.