This is an example of the sort of thing I hate. People take an extreme anecdote - sometimes one that's been peer-reviewed, sometimes not even that - and then try to present the extreme anecdotes as being "normal and characteristic of the disease". Asymotomatic transfer? Yes, it can happen, no it's not normal and characteristic. Spread by toilet aerosols (fecal-oral route)? Yes, it could happen, no it's not normal and characteristic. Relapse? Yes it can happen; no it's not normal and characteristic. Infected by a contaminated surface a week after it was contaminated? Yes, it could happen; no, it's not normal and characteristic. Infected twice? Theoretically possible, not normal and characteristic, and probably never happened in the first place. On and on.
Yes, it's "possible" to be infected from 5 meters away. No, that's not normal and characteristic.
And this is the sort of stuff that's been bothering Musk too, and I fully, 100% sympathize.
And back to the initial question: you don't turn to armchair epidemologists with a pocket calculator when discussing the CFR; you consult WHO and the CDC. And most recently they've been pointing at a CFR of about 0,7% outside of Wuhan, give or take. Musk was absolutely right to call out that sort of stupidity.