U-6's 7.2% is compared to 7.0% in February 2020 and 8.1% in February 2007. So even that is similar to pre-Covid levels and better the the peak before the last recession.
The 5% = full employment threshold has always been calculated based on U3. That's what they measured to come up with that estimate. If you want to talk about how U6 is better, then you need to compare it to previous U6 measurements, not previous U3 measurements.
If, as you were implying, U3 was hiding extra unemployment right now but U6 was more accurate, then it wouldn't be so closely tracking U3.