> Offering insurance as compensation started during WWII when wages were frozen by the government.
Absolutely true, and those regulations haven't been in effect for the better part of a century now.
If you want to argue that long-discontinued government programs can entrench systems that continue to produce market distortions today then I'm happy to agree with you, but that's the exact stated rationale for the broad suite of civil rights regulations that in my experience most libertarians oppose on economic freedom grounds. I don't think you get to have it both ways.
> There is also the fact that employer-provided insurance is almost entirely untaxed.
Again, true. But I think you're drastically underestimating how beneficial it is to an insurer to cover groups of mostly healthy people (i.e. a company's entire labor force) instead of groups of mostly sick people (i.e. those most motivated to acquire private health insurance).
In order for the market for healthcare to be truly free, in the sense you're describing, medical professionals would need to be willing to consistently refuse service to people who can't pay, which is frequently in violation of their professional ethics. Since that (entirely private and voluntary) market distortion isn't going away any time soon, insurers who can find other mechanisms to incentivize relatively healthy people to contribute to the risk pool can provide the same coverage at a lower cost per person.
That means that, tax incentive or no, insuring a broad cross section of mostly healthy people as a group is always going to be cheaper than insuring them individually. And the more healthy people can be incentivized to sign up, the cheaper it becomes for all participants, which gives whoever is making decisions on behalf of the group an incentive to encourage as much of their potential pool to participate as they can. The taxes reinforce this system, but it would still be self-perpetuating without them.