Minimally support WebAssembly Relaxed SIMD builds

Relaxed SIMD currently use the same code paths and microkernels as WAsm SIMD128

PiperOrigin-RevId: 417704438
diff --git a/bench/qs8-gemm.cc b/bench/qs8-gemm.cc
index 156818e..1aa54fb 100644
--- a/bench/qs8-gemm.cc
+++ b/bench/qs8-gemm.cc
@@ -1653,7 +1653,7 @@
 #endif  // XNN_ARCH_X86 || XNN_ARCH_X86_64
 
 
-#if XNN_ARCH_WASMSIMD
+#if XNN_ARCH_WASMSIMD || XNN_ARCH_WASMRELAXEDSIMD
   static void qs8_gemm_2x4c2__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {
     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2_ld64, 2, 4, 2, 1,
       xnn_init_qs8_conv_minmax_fp32_wasmsimd_params);
@@ -1785,7 +1785,7 @@
   BENCHMARK_GEMM(qs8_gemm_3x4c8__wasmsimd_mul16_ld128)
   BENCHMARK_GEMM(qs8_gemm_xw_2x4c8__wasmsimd_mul16)
   BENCHMARK_GEMM(qs8_gemm_xw_3x4c8__wasmsimd_mul16)
-#endif  // XNN_ARCH_WASMSIMD
+#endif  // XNN_ARCH_WASMSIMD || XNN_ARCH_WASMRELAXEDSIMD
 
 
 static void qs8_gemm_2x2__scalar(benchmark::State& state, const char* net) {