feat(poly_mul): implement synchronous PolyMul with base-case multiply

Phase 2.2: NTT-domain polynomial pointwise multiplication.
- basecase_mul.v: degree-1 base-case multiply (c0,c1) with Barrett
- poly_mul_zeta_rom.v: 128-entry zeta ROM for PolyMul
- poly_mul_sync.v: FSM (IDLE→LOAD 256 cycles→COMPUTE 256 cycles→DONE)

Verified: 5/5 vectors bit-exact vs Python PolyMul reference

test_framework/modules/poly_mul/test_plan.json

@@ -0,0 +1,21 @@
+{
+  "module": "poly_mul",
+  "rtl_top": "sync_rtl/poly_mul/poly_mul_sync.v",
+  "rtl_deps": [
+    "sync_rtl/poly_mul/basecase_mul.v",
+    "sync_rtl/poly_mul/poly_mul_zeta_rom.v",
+    "sync_rtl/ntt/barrett_mul.v"
+  ],
+  "tb_cpp": "sync_rtl/poly_mul/TB/tb_poly_mul.cpp",
+  "simulator": "verilator",
+  "timeout_s": 300,
+  "cases": [
+    {
+      "id": "basic",
+      "description": "Pointwise multiply of two random NTT-domain polynomials, vs Python basecase_mul reference",
+      "params": {},
+      "num_vectors": 5,
+      "tolerance": "bit_exact"
+    }
+  ]
+}