Default USE_FLAGTUNE to 0 and add fold_scale A/B script with trace shapes

Plain autotune is the common case (fast, same-config A/B); FlagTune's
exhaustive first-run search is now opt-in via USE_FLAGTUNE=1.
ab_fold_test.sh automates the FLAGGEMS_MXFP4_FOLDSCALE=0/1 record-replay
comparison over the four DeepSeek-V4-Flash trace shape sets, which are
now tracked under runs/shape/.

run_pytest.sh

@@ -46,8 +46,8 @@ fused_marlin_moe_mxfp4:
   shape_desc: "num_tokens, num_experts, hidden_size, intermediate_size, topk"
 YAML
 
-# 调优空间：1=FlagTune 扩展空间（首跑全量搜索、慢）；0=普通 autotune（快速验证）
-USE_FLAGTUNE="${USE_FLAGTUNE:-1}"
+# 调优空间：0=普通 autotune（默认，快速验证）；1=FlagTune 扩展空间（首跑全量搜索、慢）
+USE_FLAGTUNE="${USE_FLAGTUNE:-0}"
 
 # autotune record/replay（见 _autotune_record_plugin.py）：
 # - 空：record 模式，本次选中的 config 记录到 $OUT_DIR/autotune_records/<op>.json；