nx_lipratio_gate.nx

nx_lipratio_gate.nx -- GATE for the buildable half of aesthetictwin rung AT4: lip vermilion heights and the upper-to-lower ratio, measured from the BLENDED SURFACE rather than from the part table. THE LOAD-BEARING TOOTH IS T4, AND IT IS THE RUNG'S DONE-RULE STATED AS AN EXPERIMENT. That rule says the axis must be "read from mesh landmarks and NOT from slider positions". Those two are indistinguishable on a healthy face -- both give a plausible number -- so the only way to tell them apart is to CHANGE THE GEOMETRY AND SEE WHICH ONE MOVES. T4 halves the lip parts' vertical radius and requires the measured ratio to change. The estate's existing nx_faceprofile_gate reads lip depth as p[10*6+2], a part-table field that halving ry does not touch at all, so it would sail through T4 unchanged -- which is precisely the defect this rung exists to close. T3 is the other half of the same idea from the opposite direction: the stomion this gate EXTRACTED from the surface must agree with the parameter that CARVED it, within a tolerance DERIVED from that carve's own radius plus the blend width rather than typed as a constant. Extraction and parameter are two independent routes to one landmark, so their agreement is a free audit and their disagreement would locate a real defect in whichever is wrong. T5 is the negative control and it is REACHABLE, not decorative: shrunk to a tenth, the face no longer intersects the midline band at all, every probe returns no-surface, and the measure must report UNMEASURED rather than a number. A measure that answered anyway would be reading its own parameters.