I think most meteorologists use the EF-scale primarily to obtain an estimate of the intensity of a tornado. For most of us, we don't really use the EF-scale to assess the engineering ramifications of tornadoes -- we look at tornado ratings to get a feeling for the intensity of a tornado. Well should all know that one of the main problems with the EF-scale is that, 99.9% of the time, any assessment is based on damage (e.g. the DoDs for the DIs), which is problematic when a tornado does not hit any significant DIs (or completely destroys those that it does hit). These ratings are then used in climatology and correlation studies to examine meteorological characteristics that may help discriminate between environments that support high-end (and often high-impact) tornadoes from those that do not. There have been some very good publications from SPC mets focused on just this; we use 0-1 km SRH, LCL height, Significant Tornado Parameter, and other parameters because they've shown utility in the past in determining when certain environments may support high-end tornado risks.
As noted previously, the problem with lack of DIs for some tornadoes can be problematic. If a tornado with 250 mph winds at 10 m AGL spins in a field and hits absolutely nothing, there's a good chance that the tornado will be underrated in the eyes of most mets (again, I think many of us work on the premise that the EF scale is proxy for intensity ). This, in turn, certainly can cause issues with correlation studies (e.g. "look, this tornado occurred in an environment of 5000 j/kg MLCAPE and 400 m2/s2 0-1km SRH, yet it was only an EF1"). If we have high-quality measurements near the surface that are convincing, then I see little reason why those data cannot be used with non-traditional indicators to augment a rating. In the case of the El Reno tornado, there was a direct surface observation within the tornado. If that tornado hadn't hit anything in the way of high-end DIs, then should that surface ob be ignored? Ideally, the EF-scale may be based on damage, but isn't the practical use of it to use damage as a proxy for intensity? If so, then any data that can be corroborated with other, non-traditional indications should be included, IMO. Heck, even without ANY quantitative measurements, there can be a considerable amount of subjectivity in the ratings. For example, some of the homes that were "slabbed" were deemed to have substandard construction. However, the result is that one can only establish a lower bound to the winds (they were at least 1xx). In addition, the majority of the the winds associated with the different DoDs for each DI are subjective, based largely on the consensus of a group of wind damage and engineering experts. The original EF-scale document lists, if I recall correctly, the range of the wind speed estimates given by the different people involved in the development of the EF-scale for each DoD and DI, and you can see that some DoDs for some DIs have a very significant range.
In this case, the winds we sampled were not 205 mph. The winds measured (~65 m AGL) were well over the lower EF5 boundary. Even a very conservative estimate yielded a sfc-estimated wind above 200 mph, with the peak radial velocity, again, very much in excess of that speed. I'd agree with being more skeptical in the use of radar data if the data were farther above ground level or much nearer the EF4/EF5 threshold, though.
EDIT: This is entirely my personal opinion and does not represent the feelings of the group with which I work.