I don't agree with this perspective. A tax on negative externalities doesn't have to be regressive. It depends on what the tax money is spent on. This is an extreme example, but if you added a congestion tax and then spent the money on a tiny UBI, you might generate $10/person/month, which would be a major uplift to the poorest in our society who don't drive at all. The argument against congestion pricing is further weakened by the fact that those harmed (drivers, pay the tolls) are also those who benefit (drivers, who enjoy less congestion). The ones who are harmed the most are those displaced from driving, who have to find something else to do and don't enjoy the benefits of reduced convention. That's using congestion pricing as an example, but the same argument applies to taxing vehicles in proportion to the wear they impose on roads.

Business owners who pay the tax are free to raise their prices, which is how it's supposed to work. They're currently raising their prices because their drivers waste time in congested traffic and because they pay taxes to the government for road maintenance.

For an analogy, it also makes sense to tax companies who dump their waste in rivers, to the extent that their waste dirties the rivers. If there is some ultra-valuable product that could only be made by dirtying a river (idk, let's say that for some reason insulin had to be made that way), it would be a good that it could still be made, while discouraging people from dirtying rivers for little reason. No one would say "polluting the river should be free because we all use products that are made by polluting rivers." If polluting rivers were free and the government just taxed everyone to clean them up afterwards, we probably all really would use products made by polluting rivers! but that doesn't mean we would be worse off by taxing it directly.

That said, I agree that there's no reason for tolls to fund the road system. Hypothecated taxes are generally not a good idea, despite the fact that they're very intuitively appealing.