I can see a few things driving this. First, I think that pretty much everybody was under more stress in that period -- anybody on the road would be less patient and less generous / more aggressive. Second, there's some sampling bias -- the people on the road would be either essential workers (underpaid, overworked, even more stressed) or oversampled from a risk-tolerant subset of the population (more inclined to see an open road as an opportunity to drive recklessly). The risk calculation of driving drunk would go down with emptier roads, too.
I think it's counterintuitive unless you account for the fact that it's intuitive to think that the roads will be empty, which should impact people's behavior. I don't think I would have bet either way, but post-facto justification seems pretty tidy.