[Deprecation] Remove deprecated plugin and compilation fields for v0.13 release (#30396)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

tests/kernels/moe/test_ocp_mx_moe.py

@@ -70,12 +70,12 @@ def test_mxfp4_loading_and_execution_moe(vllm_runner, model_case: ModelCase):
             f"{torch.cuda.device_count()}"
         )
 
-    # `cuda_graph_sizes=[16]` to reduce load time.
+    # `cudagraph_capture_sizes=[16]` to reduce load time.
     with vllm_runner(
         model_case.model_id,
         tensor_parallel_size=model_case.tp,
         load_format="dummy",
-        cuda_graph_sizes=[16],
+        cudagraph_capture_sizes=[16],
     ) as llm:
         # Disabled as check_model is broken: https://github.com/vllm-project/vllm/pull/18465#issuecomment-3329880562
         # def check_model(model):