tofu currently implements the rocking assuming specular reflections around bragg's angle
But the way the rocking curve is actually measured (fixed source and detector + rocking crystal) actually suggests it should instead be used otherwise, with the reflected ray at a complementary angle to the incident angle with respect to Bragg's angle.
Context: