Fix qd8_f16_qb4w GEMM config getter always returning NULL on Arm

xnn_init_qd8_f16_qb4w_gemm_config() gates its return on
qd8_f16_qb4w_gemm_config.arch, but init_qd8_f16_qb4w_gemm_config() never
assigns .arch on any Arm branch (only mr/nr/log2_kr/planes/dqgemm). So on
Arm64 the getter always returns NULL, and every QD8/F16/QB4W fully-connected
op fails at runtime creation with xnn_status_unsupported_hardware. This
surfaces once the convert->FC packed-LHS path sets INLINE_LHS_PACKING and
setup_variant_and_gemm_config gets a NULL gemm config.

Align with the qd8_f16_qc4w sibling getter, which returns the config
unconditionally (its init likewise does not set .arch).

Verified on Apple M4: a QD8/F16/QB4W linear .pte that previously aborted
with unsupported_hardware now runs (Model executed successfully).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: notaJiminLee

src/configs/gemm-config.c

@@ -6275,7 +6275,7 @@ const struct xnn_gemm_config* xnn_init_qd8_f16_qb4w_gemm_config() {
     return NULL;
   }
   XNN_INIT_ONCE(qd8_f16_qb4w_gemm);
-  return qd8_f16_qb4w_gemm_config.arch ? &qd8_f16_qb4w_gemm_config : NULL;
+  return &qd8_f16_qb4w_gemm_config;
 }
 
 const struct xnn_gemm_config* xnn_init_qd8_f32_qc4w_gemm_config() {