[CI][XPU] Disable model runner V2 for XPU quantization test for some partially pre-quantized models (#56179)

Signed-off-by: Chaojun Zhang <chaojun.zhang@intel.com>

.buildkite/intel_jobs/test-intel.yaml

@@ -145,10 +145,16 @@ steps:
     source_file_dependencies:
       - vllm/
       - .buildkite/intel_jobs/test-intel.yaml 
+    # TODO: XPU fused MoE + model runner V2 currently hangs on some
+    # pre-quantized models, e.g.
+    # quantization/test_online.py::test_online_quantization_loads_real_weights[nm-testing/tinysmokeqwen3moe-W4A16-first-only-CTstable]
+    # Remove the VLLM_USE_V2_MODEL_RUNNER=0 override below once the hang is
+    # root-caused and fixed.
     commands:
       - >-
         bash .buildkite/scripts/hardware_ci/run-intel-test.sh
         'cd tests &&
+        export VLLM_USE_V2_MODEL_RUNNER=0 &&
         pytest -v -s quantization/test_auto_round.py &&
         pytest -v -s quantization/test_online.py'
   - label: "XPU compressed tensors FP8 test"