There are no reliable estimates of overall fatality of the virus. Only the estimates of how many people _who required treatment_ have died, which the press deliberately misinterprets as the overall fatality rates most of the time, because higher numbers drive more clicks. Watch epidemiologists speak about this: they will go out of their way to point out that we don't know these percentages, and we won't know them for quite some time. Your "numbers" are bogus. You can easily see this if you look at the _initial_ estimates of H1N1, SARS and MERS. While COVID19 does appear to be deadlier, _nobody_ knows what the fatality rates will look like population-wide. Anyone who claims otherwise is full of shit. What we do know is that the numbers we currently have are the very uppermost bound of what's possible.

Nope, SARS and MERS are way deadlier than Covid-19, not just in initial estimates. South Korea tested hundreds of thousands of people and we have the Diamond Princess. That makes it unlikely that we are overestimating CFR by an order of magnitude due to uncounted mild cases. If you have someone reputable who thinks true CFR could be as low as 0.1%, please share! As far as I can tell no one believes it can be significantly below 1% for a country with similar demographics as the UK.

The reason why Covid-19 is scarier than SARS is precisely because it has a lower fatality rate, allowing it to spread instead of killing its host before they can spread it.

Not true. These estimates are of overall fatality based on number of people diagnoised. Not based on those who required intensive treatment. That means we have crude estimate on death rate for people showing some symptoms. Crude in a way not everyone in the pool is recovered, or dead for that matter, so it's not possible to be that accurate yet, but it will only be higher not lower.

Wouldn’t we only have a crude estimate on the death rate for people diagnosed with the virus? If we find that many people are not getting tested for the virus and just recover by themselves, then it’s possible for the fatality rate to be lower.

Well, those asymptom case can also infect others. In all likeliness they will infect others, multiple others. So if epidemiological root trauce is done right, we would find those asymptom carrier.

This doesn't mean those cases are "asymptomatic". Most people who get the disease fare about as they would with a flu. I'd also venture to guess that most people do not go to the hospital unless things are really bad (e.g. they can't breathe on their own). Furthermore, because not everyone at the hospital gets tested, I'd also guess that there's a number of cases where deaths are misattributed to plain old flu, too, since it's the flu season still, and some of those who are dying "of coronavirus" would have died regardless.

Well. How is common flu any different? Most people don't get tested and they just recover, if any, a lot more don't. So seeing those two numbers on the same page should mean something, even if you doubt the accuracy based on the assumption a large quantity of people aren't tested.