An inward wave doesn't seem likely on a tire that's anywhere near to normal inflation: that would mean the tire tread lifting off the road; but the air pressure inside is pressing the tire tread down onto the road. Even if the tire is only at 15 psi, with a contact patch of several tens of square inches that's still a lot of force. For the tire to lift off the road in the middle is plausible for a completely flat tire, which is where someone who posted to this thread observed it happening; but it's different if the tire is just low, not flat.
The nitpicker in me says that the tread will certainly compress, even if the belt doesn't, since the tread is outside the belt and is normally of a larger radius, yet as the two go over the contact patch, they are flattened out to the same length, or rather, to be more precise, to the same (infinite) radius. But this happens even with a tire at normal pressure, and happens about the same amount, since the initial radius is almost the same, and so is the final (infinite) radius.
To address your real question, the steel belt is prestressed by the tire's air pressure. For it to compress much, it'd have to lose all that prestress. Then it'd have to buckle -- which the individual wires in the belt are restrained from doing by the rubber surrounding them and binding them together. Either the rubber would have to disintegrate, or the tread as a whole would have to buckle. It can't buckle downward into the road, because the road is hard, so it would have to buckle up off the road, which gets us back to the question of whether the tire lifts up off the road in the middle of the contact patch; as mentioned above, I don't think this is likely for the sorts of tire pressures which you'd like to catch and report to the driver via an automated system of the type that started this thread.
Much as I like conservation-of-energy type arguments, they are made inapplicable here by the energy loss that heats up the tire. The actual torque at the axle is going to be greater than the number yielded by your above formula, because the torque at the axle must also provide the energy that goes into heating the tire.
The basic situation seems to be this:
-- The tire can't slip on the rim; one revolution of the tire always means one revolution of the rim.
-- At the middle of the tire contact patch, the tire wants to move the wheel at the angular speed omega given by the formula omega = v/r, where r is the distance from the axle to the road, and v is the car's speed.
-- At the very front of the tire contact patch, the tire wants to move the wheel at approximately omega = v/r0, where r0 is the distance from the axle to the front of the tire contact patch, which is about the normal radius of the wheel.
The two parts of the tire are thus fighting each other: each wants to move the wheel at a different angular velocity. The resulting angular velocity will be between the two figures, and there'll be scrubbing of the tire against the road, with the front of the contact patch trying to go faster over the ground, but slipping backwards, and the middle of the contact patch trying to go slower over the ground, but slipping forwards.
(Some of the people in this thread have been arguing for omega=v/r; others have been arguing for omega=v/r0; well, you're not the only ones; the tire is arguing with itself too.)
Sharp eyes will have noticed that there's an "approximately" in my above argument for omega=v/r0. That's largely because the tire tread is hitting the ground at an angle; that's really a formula for the velocity of a point on the tread just before it hits the ground, not just after. The correction factor for this is roughly cos(theta), where theta is the angle between vertical and the line between the axle and the front of the contact point: really the formula should be more like
omega = v*cos(theta) / r0,
except even that isn't correct, since it's based on the approximation that the tread makes a perfect circle interrupted only where it touches the ground, where it's perfectly flat. Still, it's not too far off; and for small theta, cos(theta) is approximately equal to one, yielding the approximation-to-an-approximation formula omega=v/r0.