Glen Romine
EF5
I have to differ you here. Take a look at this distribution of tornado strength between discrete vs. linear modes from Elaine Godfrey's work here. I remember the sampling of tornadoes documented by the DOWs to be skewed toward high end relative to other climatologies.
Jim
As Jeff specified, the DOW climatology is based on supercell tornadoes (largely from the Great Plains region) - I don't think they included any QLCS type tornadoes in the study, and they were suggesting that tornadoes with a parent supercell storm appeared to have a preferred intensity in the EF2 range. Whether lumping in all tornadic vortices returns the overall distribution back toward log-linear is not answerable. Still, it seems plausible there is possible under-rating of QLCS tornado damage as well given the likelihood of the tornado at max intensity not interacting with appropriate structures for a 'proper' rating. Finally, could also be some bias from the tendency for QLCS type tornadoes to be more common in more 'developed' land areas, whereas supercell tornadoes are more common in the Great Plains where land improvements are more sparse.
As for Ilya's observations, you should probably share whatever you have with the NWS. There were several SPC folks on the storm, and I gather none of them saw the stovepipe either - though they would likely have approached the storm from the east. Surely some Texas Panhandle chasers would have been in better position to have seen this earlier tornado.