[Feature] support aclgraph for model runner v2 (#7110)

### What this PR does / why we need it?
This PR aims to support aclgraph for model runner v2, please see RFC
#5208. The PR contains these modifications:
- adapt to newest commit of vllm main branch.
- supply a unified interface of extra forward context for both model
runner v1 and model runner v2.
- implement graph mode for main model. 

### Does this PR introduce _any_ user-facing change?
no

### How was this patch tested?

- vLLM version: v0.16.0
- vLLM main:
4034c3d32e

---------

Signed-off-by: Ronald1995 <ronaldautomobile@163.com>

tests/e2e/nightly/single_node/ops/singlecard_ops/test_fused_moe.py

@@ -184,7 +184,7 @@ def test_token_dispatcher_with_all_gather_quant(
 ):
     context_mock = MagicMock()
     context_mock.fused_moe_state = 0
-    with patch("vllm_ascend.ops.fused_moe.moe_mlp.get_forward_context",
+    with patch("vllm_ascend.ascend_forward_context.get_forward_context",
                return_value=context_mock):
         a = torch.randn((m, k), device=device, dtype=dtype) / 10
         w1 = torch.randn((e, k, 2 * n), device=device, dtype=torch.int8)

tests/e2e/singlecard/model_runner_v2/test_basic.py

@@ -85,3 +85,29 @@ def test_egale_spec_decoding(
         },
     ) as runner:
         runner.model.generate(prompts, sampling_params)
+
+
+@pytest.mark.parametrize("model", MODELS)
+@pytest.mark.parametrize("max_tokens", [32])
+@pytest.mark.parametrize("enforce_eager", [False])
+@pytest.mark.parametrize("compilation_config", [{"cudagraph_mode": "FULL_DECODE_ONLY"}, {}])
+@patch.dict(os.environ, {"VLLM_USE_V2_MODEL_RUNNER": "1"})
+def test_qwen3_dense_graph_mode(
+    model: str,
+    max_tokens: int,
+    enforce_eager: bool,
+) -> None:
+    prompts = [
+        "Hello, my name is",
+        "The president of the United States is",
+        "The capital of France is",
+        "The future of AI is",
+    ]
+
+    sampling_params = SamplingParams(max_tokens=max_tokens, temperature=0.0)
+    with VllmRunner(
+        model,
+        max_model_len=1024,
+        enforce_eager=enforce_eager,
+    ) as runner:
+        runner.model.generate(prompts, sampling_params)