This happened in 2012 as well. I did not pay attention to 2016, so I can’t say whether something similar happened then.
First, people are irrational. Second, on heavily-scrutinized lines, bookmakers generally want to try to balance the amount of money bet on either side, so as to make money from the vig, rather than hoping for one outcome or the other. The likely reason why the odds have been shortening is because the Trump betting has been high, particularly when the odds have been, as you said, around 2/1.
There’s also rational reasons why the odds should be shortening. For example, it’s clear that a lot of mail-in voting favors Biden. However, unlike voting at the polls, there’s a decent chance that a fraction of those votes will be invalidated due to not following the instructions properly or because signatures by some judgment do not match. Thus the polls favoring Biden likely overestimate his ultimate vote share. For example, if the aggregate of polls suggest he will receive 60% of the vote, it may turn out to just be 57% instead, due to this effect.
That being said, people are certainly irrational. So many people misunderstand the lessons of 2016. Pollsters are not stupid; the ones that called 2016 wrong re-assessed and resolved to do better. They changed their models, and did a superb job calling 2018. So many comments in this subreddit discount polls just because they did not perform as well as they should have in 2016. Polling is an imperfect science, but for predictive power it easily beats the half-baked prognostications I see daily.
Ultimately, all these comments are worthless. Whatever happens in 2020 will happen, and people who wager on this upcoming election will either win or lose. The only thing, if you want to bet on this, is to try to approach the wager not as a political partisan but as a person who seeks to maximize one’s return. Personally, I don’t care about any ginned-up October surprise except how it affects the polls, because the polls are our instruments of assessing the will of the electorate and therefore predictive of what will happen in the election.