I didn't bother addressing the graph you presented because it doesn't present any information. It doesn't say what it's based on. It doesn't say what data it is presenting. It doesn't where any of its data came from. Was it national? state data? county data? Who knows? There's simply no basis for putting any confidence in the wiggles on that graph.
I can agree that some people would lie to exit pollsters. But it makes more sense to simply refuse to answer any questions. After all, lying is a sin. Lying degrades you. People don't lie for fun. They lie when they believe that a lie is morally necessary. Such is not the case with exit polls. Hence, I find your hypothesis that millions of people lied to the pollsters to be not credible. Indeed, I think that a more plausible hypothesis is that voting computers were hacked to alter the results. That's also an incredible hypothesis--but it is more credible than the "people lying" hypothesis.
Moreover, there are many reasons to conclude without any knowledge of polling results that Mr. Biden will defeat Mr. Trump. We've accumulated lots of knowledge about the impact of various factors on the outcome of an election. Incumbency, as you pointed out, is an advantage. But the single most important factor is the state of the economy, in particular the unemployment rate. The current economy is in spectacularly bad shape; the only comparable economic situation was the Depression. Unemployment is at record levels. Throw in the pandemic with 170,000 dead Americans and no end in sight, and ANY Democrat would defeat Mr. Trump.
Yes, there could be something to reverse the situation. Perhaps Mr. Trump will single-handedly produce a vaccine. Perhaps Mr. Biden will be caught in bed raping orphans. But barring some dramatic reversal, the experiment in November is not likely to surprise anybody but die-hard advocates of Mr. Trump -- who will surely claim fraud.