This market exists for two main reasons: I think a cumulative-independent approach is the most useful format for markets like this, and I also wanted a market where people could directly compare their predictions about the popular vote margin to previous presidential election results.
Dave Leip's numbers are used and will be used to resolve this market.
Note that if it gets close, the answers that refer to a specific margin should be taken as not the exact number listed in the answer, but the actual percentage of the margin of the popular vote in that election (e.g. the 2020 result should be considered 81286454-74225926=7060528, 7060528/158594895= ~4.44519264%) according to Dave Leip's numbers.
I've made Dave Leip's numbers the resolution criteria this time, so you won't have to wait for the FEC report. If Dave Leip's website explodes I'll use whatever looks like consensus to me among other election results sources, in the interest of speedy resolution. If there's a big disagreement between reputable sources for specific thresholds, then I'll wait for the FEC report for those specific threshold answers and resolve the rest.
Please ask any questions in the comments if you have any.