There are a couple of problems with your reasoning. First, according to your chart, there should be about 15-16 million more registered voters than in 2018, if earlier trends continue. Voter registration went up about 16 million from 2010 to 2012, and around 15 million from 2014 to 2016. Second, following that chart assumes that circumstances and voter registration trends did continue as before without increasing. As I've mentioned a few times, there were reasons to expect a greater than usual increase in voter registration (and voter turnout) this election cycle.
As to sources, if you look back at my link to registered voters (
Number Of Registered Voters by State 2020) you will see that they cite various state websites for their numbers, for example
www.sos.wa.gov for Washington, and elections.delaware.gov for Delaware. The other source I used, for total population and voter-age population, was the US Census.
To go back to Statista, according to them, overall voter turnout in the 2020 election was 66.7%
2020 Presidential Election: voter turnout rate U.S. 2020 | Statista So if you are going to use Statista to claim near 100% turnout, you should probably check to see what they actually have to say about voter turnout.
Finally, as best as I can determine in a quick check, Statista has been around since 2008 while World Population Review has been around since 2011. The US Census has obviously been around longer. So the argument that Statista would not risk their reputation can apply equally to the sources I've used. Additionally, Statista appears to be a German owned site, which is pretty funny considering the rumors that have been going around about a raid in Germany to get servers holding evidence of election fraud.
Again, none of this disproves your claims, but it does make your argument a lot less compelling. It doesn't help that Statista keeps its sources behind a paywall.