nx_skin_ita.nx

nx_skin_ita.nx -- SKIN ALBEDO DERIVED FROM THE GENOME, VIA A PUBLISHED STANDARD (charsim R0, debt 1786549493 residue a). LIB, no main (ecosystem convention). WHAT IT REPLACES AND WHY. nx_nxa_texbake_lib baked skin from TWELVE free sRGB constants (NTB_FACE_R=228, NTB_TORS_G=184, ...). Nothing derived them, nothing could falsify them, and every bred creature got the identical skin -- a fixed ramp wearing the shape of a material model. Rule 25 says build intelligence, never fewer options; rule 11 says a threshold nobody can point to a source for is a defect. This organ removes all twelve. THE STANDARD. Dermatology classifies constitutive skin colour by the INDIVIDUAL TYPOLOGY ANGLE (Chardon et al. 1991), defined in CIE L*a*b* as ITA = atan((L* - 50) / b*) in degrees, with six published categories separated at 55, 41, 28, 10 and -30 degrees. Those five boundaries are the ONLY classification numbers in this file, and each is stored as its TANGENT so the whole organ runs in integers with no trigonometry at all: comparing ITA against an angle is exactly comparing (L* - 50) / b* against that angle's tangent. WHAT IS DERIVED vs WHAT IS CHOSEN -- stated plainly, because a reader who cannot tell will trust the chosen ones as measured (the fabricated-constant class): DERIVED from the standard: the five class boundaries; the lightness endpoints of the locus, which fall out of the boundary tangents rather than being picked (see si_lstar_of_gene). CHOSEN, and this is a MODEL not a measurement: that skin lies on a straight locus of fixed a* and b* (literature-typical redness and yellowness), and the four per-region lightness offsets. They are expressed as offsets in L*, a perceptually uniform axis, so "slightly lighter" means the same thing everywhere -- which is precisely what twelve free sRGB numbers could not express. The IMPRECISION accepted: fixed-point, with the fifth root taken at 1e3 so one output step is under half an 8-bit code. Colour is exact to the byte after the final search. NO SECOND TRANSFER CURVE. si_srgb_of_lab finds the sRGB byte by searching the FORWARD IEC 61966-2-1 transfer, so this organ carries the generating function, not a rival copy of nx_eink_tone's baked table. The gate proves they agree on that organ's own banked anchor (sRGB 128 -> linear 21586 -> L* 53.59); consolidating both onto one shared lib is filed, not done here. license_tier: ORIGINAL No hw writes (Rule 26).