The leverage on the chain that goes from the top of the chainring back to the cogs changes as the chain engages at different points on a non-circular chain-ring, including on oval one. This is a fact, and simple high school level physics.

That Wiki-page you linked to actually explains this. And you see the concept put into practice all around you. Brake levers that change the rate of cable pull throughout the stroke use this concept. So do many workout weight machines that use belts going over pulleys that are not circular.

I am afraid I can't quite follow where you are going amiss here, but I do notice a few things you seem confused about:

You are getting distracted by the fact that the same number of teeth are engaging the chainring. That is a red herring. Think about it: in the way you describe this is also true (well, sort of, see next paragraph) of the non-circular gear on the Wiki page you linked to and refereed to, yet you accept the fact that the leverage changes in that instance. The mechanical advantage of a gear has zero to do with how many links are engaged.

Also, while it is not even relevant, I should point at that the chain is NOT always over half of ANY chainring (round or otherwise). It depends on the size of the cog in the back, and the orientation of the tension pulley on the rear der. (these two factors change the angle at which the chain meets the chainring, and therefor the location on the ring). And as the oval ring is oriented with the long orientation horizontal, it will have less chain engaged compared to when it is in a vertical orientation (again, because of the angle of the chain). The difference is small, but it is there. BUT..... this is all irrelevant, and you may want to examine why you thought it WAS relevant, as that might be a clue as to where your reasoning is going awry..

The issue of whether the tension on the RD cage changes is also a red-herring. It is irrelevant.

Also, there is no "overall radius" in an oval. There is one for the long axis, and one for the short axis, and theoretically an infinite number in between. You are confusing radius with circumference. It is true that the circumference of a 32t round and 32t oval ring are essentially the same. But that is not what determines leverage of a gear on a chain (though in a round gear the circumference and radius always change together in a linear fashion).

The leverage (mechanical advantage) that a gear has is determined by it's radius, not it's circumference. this wiki page talks about this https://en.wikipedia.org/wiki/Mechanical_advantage

Circumference vs radius is a very important distinction here, because it is the RADIUS that determines mechanical advantage. Specifically, the radius where the chain is being picked up.

It can be a little confusing because cyclists always talk about tooth count of sprockets (which is a measure of circumference) when discussing gear ratios, but the reason this works is because for round rings, the radius is consistent, and varies linearly with circumference.

Also, you keep confusing "power" and "leverage". Oval rings DO vary the leverage throughout the rotation. That is an objective fact, backed by simply physics. And subjectively, you definitely notice it. Whether they do anything in terms of your power output is a bio-mechanical question and a very complex one at that, and whether they actually HELP you is subjective.

Good luck getting a handle on this, but I can give you some advice, you will save yourself a lot of time if you focus on what you are not understanding correctly, rather than trying to convince others that the basic laws of physics don't apply here.

What IS true is that the the OVERALL gearing throughout the entire crank revolution is the same in the sense that you go just as for in one crank revolution with a 32t round or 32t oval ring. What changes is the leverage throughout the crank revolution. At some points it is higher than a round, and at others lower.