nx_ice_stabiliser_gate.nx

nx_ice_stabiliser_gate.nx -- gate for the hydrocolloid term. T0 ★LBG is INERT without milk protein and works with it (310 permil) T1 the protein-conditional flag separates LBG from xanthan T2 ⚠GUAR -- the gum the reference product uses -- has NO measured protein-conditional data in either direction T3 gelatin's illegible cell reads UNKNOWN, never a substituted guess T4 the CONTROL ITSELF MOVES with protein (4.02 -> 4.59), so retardation must be taken against the matching control T5 saturation at 0.3% w/w: dose past it is wasted, and measured as such T6 recommended stabiliser FALLS as fat rises (35 -> 25 -> 15 per 10k) T7 carrageenan minimum is dosed off PROTEIN: 13 per 100k at 41 permil T8 carrageenan window is bounded BOTH sides: under-dose and gelling T9 incompatibility xanthan > guar > LBG; guar demands a secondary T10 ★★THE TRAP: Flores means show a 258 permil apparent gain and the study found NO significant difference -- both are returned T11 ★the refusals: model result does not transfer; gum swap unsupported T12 mechanism is ice adsorption, not viscosity; storage, not freezing T13 unknown gums and impossible doses REFUSE license_tier: ORIGINAL expect_exit: 0