And that's counting people who have bothered to get tested. If younger people are getting asymptomatic or minor symptom covid that looks like a cold, then the "true" ifr will be even lower

Infection fatality rates are based on antibody seroprevalence studies and thus accounts for asymptomatic cases. What you're thinking of is the case fatality rate, not the infection fatality rate.

You can't get an accurate community seroprevalence measure when your community is told to shelter in place.

Seroprevalence studies have been conducted all over the world. For example in Delhi India approximately 50-60% of the population now have antibodies for SARS-CoV-2. There are statistical techniques for adjusting the raw results based on population demographics.

https://www.thehindu.com/news/cities/Delhi/delhi-may-be-head...

Note that seroprevalence studies will underestimate the number of infections because some recovered patients don't produce detectible levels of antibodies.