Aren't more highly educated voters also more likely to vote, though? So perhaps it balances out to some extent.
Not weighting by education is one of the reasons why polls and exit polls in 2016 were found after the fact to have undercounted WWC voters.
Yes, but this year is very likely to be a different electorate. I think it's reasonable to assume that in 2016 lower educated voters were more motivated (relative to previous elections) than higher educated voters were, but in 2018 the reverse is probably true.
This is a reasonable expectation and I share it, but that doesn't mean it's a good idea not to poll education at all.