AFAICT, that 0.03% is for the U.K, not the U.S. And it doesn't say what the definition of CFR was. I've read a significant number of papers (primary materials) about SARS, MERS, H1N1, and influenza over the past few weeks. IME, CFR is usually defined as diagnosed or symptomatic cases as the denominator (which, as I've said, was until recently effectively equivalent), especially mid pandemic. But it's not always so defined. One notable exception that adds to the confusion is seasonal influenza, where U.S. annual fatality rates are calculated using post hoc modeling.

Even if we pin down CFR to be # deaths / # confirmed symptomatic, there's obviously still going to be some variability, possibly significant--quality of medical interventions, age distribution, etc. But trying to shoehorn IFR and other definitions into that number would compound such problems many fold.