Looking at the final polls in 2016, they're really not so far off. Clinton won the popular vote by 2 points. The final polls, with sampling ending within a few days of the election, showed a relatively wide range, from Trump winning by 5 points to Clinton winning by 7. The majority of the polls were within the 3-5 point lead for Clinton range. So they were only a couple of points out in reality, and they were closing, indicating that the email stuff was taking an effect.
wiki list of polls
here
I think the accepted truth that the pollsters were way out in 2016 is overstated. They were somewhat out. Guardian poll tracker average still gives Biden an 8.5 lead. And there is no last-minute equivalent of the 2016 emails. They really will need to be way out this time for Trump to have any hope.
The most striking pattern of the 2016 polls is that they overestimated how many people would vote for a third candidate. That's the bit they got most wrong. Every poll gave Trump a smaller share than he actually got, and all but one gave Clinton a smaller share. Every single one gave third candidates more votes than they actually got.