I have been following this project because I, too, have built refractors. Testing refractors is different from testing mirrors because the design of a refractor objective corrects chromatic aberration (as well as an achromat can), spherical aberration, and coma. This relies on the interplay between both the elements, the surfaces, the thicknesses, refractive indices, and spacing. For best results, it is not a willy nilly affair fixing a hole in one surface with a hill in another.
For best results, the radii have to be close to the design intent as well as the surface figures spherical. The recommended tolerance for R2 and R3 should be within +/- 1/20th % of their design value. R1 about 1/10th % and R4 fairly loose. Keep in mind there are three aberrations being corrected.
Autocollimation testing ONLY verifies that the spherical aberration is corrected, but the color and coma is in the radii.
for best results, a spherometer is not adequate to validate that the radius is correct. There should be some more precise method. Often used is measuring R3 with a knife edge and carefully measured rod. The results from this method should agree with calculation based on a spherometer by only a few thousandths.
Attached is a picture showing measuring R3 of one of my flint elements with rods.