Even the betting sites where Trump is extremely (as in, much more than in the poll-based forecast sites) favored to win the electoral college, like Polymarket, have Harris a strong favorite to win the most votes, just as the poll-based forecasts generally do.

E.g, Polymarket, which gives Trump a 65% chance to win the EC compared to 538s 53%, has Harris at 58% to win the popular vote compared to 538s 65%. (And while it is possible for the popular vote winner to have a mere plurality of votes, it is not possible to get a majority of votes and not be the popular vote winner, so, no, the source of evidence you cite does not support the claim that "by the looks of it, the majority of people" will vote are going to vote for Trump, but instead support that the majority will most likely not do so.)