Ok, I distill from your answers that if I want to take a measure of the ratio, I can use method C (separate measure with the dome at each side of the object, pointed at the camera) and it will work just as if I had used method B (separate measure with the flat thingie pointed toward the light source).
I can ignore the existence of the flat thingie, and live happy with my incident meter which only has the domed thingie.
PS Stillsilver wanted to give me this answer I guess but inverted method C and method B. Otherwise it is clear. I'll do as Mark Barendt says. I just use method A for average exposure, and C for checking ratios, and that's all. And yes, I can skew final exposure toward one of the two readings with method C if I want to expose in a non-average way.