nx_simd_sse_kat_gate.nx

nx_simd_sse_kat_gate.nx -- SOVEREIGN 128-bit SSE2 vector run-KAT: the FIRST rung of the sovereign VECTOR backend, proving xmm load/compute/store on REAL x86 silicon (NO gcc / NO qemu / NO binutils). AUTHOR=ORGAN, no-false-green. What it proves: the sovereign encoder+assembler now emit CORRECT 128-bit vector MEMORY load/store (nxasm_x86_enc.x86_sse_mem + the nxasm_x86 movdqu/movdqa K_MEM dispatch). It ASSEMBLES two tiny programs with the team's own _offc/nxasm_x86_main.elf and EXECUTES the resulting ELFs: POS: movdqu-load vecA=[10,0,0,0] + vecB=[32,0,0,0] (4x u32 LE); paddd -> lane0 = 42; movdqu-store; reload lane0 -> exit(42). NEG: identical, then pxor %xmm0,%xmm0 (zero) -> store -> exit(0). GREEN iff pos==42 AND neg==0 AND pos!=neg. Distinct pos/neg => the gate reads the ACTUAL silicon-computed exit, never a baked constant. This run- proves FOUR encoder forms end to end: movdqu-load(NEW) + movdqu-store(NEW) + paddd + pxor -- the latter two were only ever EMITTED (SHA-NI core), never EXECUTED until now. Foundation for the 256-bit(VEX) / 512-bit(EVEX) rungs. Writes knowledge/status/simd_vec_kat.log. license_tier: ORIGINAL