[Feat][Graph] Support DeepSeek with ACL Graph (#2707)

### What this PR does / why we need it?
In memory of #677 , a long overdue milestone. Now DeepSeek V3/R1 should
be OK with ACL Graph.

### Does this PR introduce _any_ user-facing change?
None.

### How was this patch tested?
Working on it.

- vLLM version: v0.10.2
- vLLM main:
68dbde5dbb

---------

Signed-off-by: Yizhou Liu <liu_yizhou@outlook.com>

tests/ut/models/test_deepseek_v2.py

@@ -41,9 +41,10 @@ def test_row_parallel_linear(cls, mock_distributed):
     assert output[0].shape == (2, 4, 64)
 
 
+@patch("torch.ops.vllm.mla_forward")
 @patch("torch_npu.npu_rms_norm")
-def test_custom_deepseek_v2_mla_attention(mock_rms_norm, mock_distributed,
-                                          base_config):
+def test_custom_deepseek_v2_mla_attention(mock_rms_norm, mock_mla_forward,
+                                          mock_distributed, base_config):
     mock_rms_norm.return_value = (torch.randn(2, 128), torch.randn(2, 128))
 
     attn = CustomDeepseekV2MLAAttention(config=base_config,
@@ -64,8 +65,8 @@ def test_custom_deepseek_v2_mla_attention(mock_rms_norm, mock_distributed,
     with patch.object(attn.mla_attn,
                       "__call__",
                       return_value=torch.randn(2, 4, 128)):
-        with pytest.raises(AssertionError):
-            attn(positions, x)
+        attn(positions, x)
+        mock_mla_forward.assert_called_once()
 
     attn = CustomDeepseekV2MLAAttention(config=base_config,
                                         hidden_size=128,