nx_identity_space_gate.nx

nx_identity_space_gate.nx -- THE REFEREE FOR THE CORRELATED IDENTITY DRAW (graphics GR24, clauses 1-3). GR24's done-rule, quoted from graphics.plan so the bar cannot drift: "semantic master parameters ... fan out into correlated low-level axes through a FULL COVARIANCE draw rather than independent per-channel sampling ... with each axis drawing from its own split stream so adding a parameter never reshuffles existing genomes; every character reproducible and auditable from its genome, proven by bit-identical regeneration from the seed and by the referee panel showing identity spread across a population without any member falling below the admitted tier floor" This gate proves the FULL COVARIANCE DRAW, the SPLIT STREAMS and the BIT-IDENTICAL REGENERATION. The referee-panel clause is NOT claimed: that panel's only voting lens is currently self-refusing, which is a blocker on the panel and not on this draw, and asserting it here would be borrowing another rung's evidence. THE TWO LOAD-BEARING TEETH: * SPLIT STREAMS -- drawing 3 axes and 5 axes from the SAME seed must give IDENTICAL values for axes 0..2. Under a single shared stream they would differ, and every genome in the back catalogue would silently change the day someone appended a parameter. Nothing else in this gate can catch that. * CORRELATED vs INDEPENDENT -- the empirical off-diagonal of a correlated population must land near its target while the same machinery with a DIAGONAL factor lands near zero. A covariance check that only looked at the diagonal would pass the exact defect this rung exists to replace. license_tier: ORIGINAL No hw writes (Rule 26).