Ad-blocking relies on knowing the origin of ads and blocking requests to the servers browser-side.
Regardless of where you shift the content it is just a matter of time until it is fingerprinted and blocked.
On the side of sites selling ads and hosting it co-mingled with other content, it is an issue of business priorities.
I know some sites that do this which causes the ads to be a feature rather than a bug. The ads are helpful and targeted specifically, requiring no fingerprinting of the user or tracking. The ad buyers get great throughput, the ad sellers get easy cash.
The model you described works fine for a directly sold placement; the publisher places the ad for a fixed amount of time for a fixed fee. They may still want to have impression and conversion tracking via a 3rd party to measure performance across the multiple sites/ad networks they may be buying on. A pure HTML ad could absolutely work and be ad-block proof, but that doesn't really scale for pubs or advertisers. I have seen that model where the publisher represents a very specific audience, like a trade association. It really doesn't work on a general content site unless there is a high degree of faith from the publisher to the advertiser.
My answer is that it’s human nature to be lazy and want to use an already existing ad ecosystem. Plus lots of systems are built around it.
And that ad blockers aren’t enough of a problem to justify leaving the ecosystem.
It’ll cost more bandwidth and hurt the site experience, but I’s imagine that’d fool adblockers.