I think "load" here is subjective as well eg a spectrum from time to first paint and a page being visible and/or interactive, up until completely loading all async and deferred content, some of which might only happen after the user crosses the fold.

I think there's a lot of interest in optimizing that initial time but less so for the full load.