You are confusing a lot of different lines of argument, and in the end I'm not even sure what you are arguing against. I think you are mostly agreeing with the proposed solution by echelon?
You mentioned AB1856 which seems waaaaay broader than emitting an age bracket header based on user settings. It puts the onus on the website operator to not only prevent presenting content to wrong age bracket users, but also to determine the age bracket of the user.
Websites are shutting down wholesale because they cannot reasonably afford the CYA, or dont want to out of principle.
In echelons scheme the parent would be voluntarily setting the age bracket on the childs device right, so if a 12 year old is more mature, then go ahead and set their device to emit the 13-16 age bracket header. If you as a parent dont believe in this, then leave the bracket unset.
For a website operator it would be trivial to block the user from accessing the site if the suggested age bracket is too low (as long as we can agree on a single way of doing things, of course). Larger operators can do more heavy content moderation and present a filtered view to those same age bracketed users.
It is true you are adding more tracking signals, and I am sensitive to the free speech issues, but children are not fully emancipated members of society yet and parents need tools to deal with the difficulties of raising children in a digital society. The alternative now seems to be OSA-like, which is even more intrusive and a risk to privacy and perhaps free society as a whole.
Of course, OSA is really the goal and not the method, and we have to remember it is never about the children. Would children have been protected from e.g. andrew mountbatten if OSA had been around at that time?