You are still measuring the circumference when you measure rollout, but in those cases it is no longer a circle so it doesn't have a diameter. Unless there is slippage, shrinkage or stretching going on, the circumference/rollout shouldn't change - and it should be equal to what you would get if you cut the tire and then flattened it out. FWIW, inflating a tire should stretch it out.

In your example, the circumference along the tread would have had to shrink for that to work. I don't know of any mechanism that could give that result. If you told me that the tire got bigger, I could actually understand that since the deformation would force more air into the rest of the tire, creating higher effective pressure and possibly stretching it a bit.

What's confusing people is that circumference is a measurement that applies to shapes other than a circle. However, it is only in the case of a circle that you can calculate the circumference using diameter (or radius) and a constant. When a tire deflects, it becomes less circular. At some point, radius (or diameter) become useless, even for estimating what the circumference is. The shape is complex enough that equations to calculate the circumference from easily observable measurements would be interesting.

However, the circumference doesn't necessarily change just because the shape changed, unless there is shrinkage or stretching. And, by definition, circumference equals rollout unless there is slippage going on.

JB