The OP asked for a way to accurately measure the radius of a fillet. The method that I proposed in post #9 will determine the radius to within a few thousandths.
There is an instrument used in optics called a spherometer which will measure the radius of a spherical surface. It consists of three points on an equilateral triangle and a fourth adjustable point at the center of the triangle.. The instrument is placed on the surface of the sphere with a ll four points touching. The distance the fourth point is above the plane of the three points determines the radius of the sphere.
I made a 2D version of this which will measure the radius of a cylindrical curve. It is essentially two points a known distance apart with a third adjustable point midway between the first two. Like the spherometer, the third point is adjusted so all three points are touching and the distance of the third point above the line between the first two points determines the radius. This can be used for both positive and negative radii.
View attachment 452904
It is much like a depth mike except the bar is replaced by the two points.
The radius of a curved surface, r, is given by the distance of the midpoint, h, above the line between the two outpoints, separated by a distance, d as r = (d^2)/8h +h/2. This device is particularly useful for large radii where only a small portion of the curve is available. For more accurate measurements, the screw would be replaced with a micrometer barrel.