[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/ut/attention/test_attention_cp.py

@@ -95,7 +95,7 @@ class TestAscendAttentionCPImpl(TestBase):
     @patch('torch_npu.npu_attention_update')
     @patch("torch_npu.npu_fused_infer_attention_score")
     @patch(
-        'vllm_ascend.attention.context_parallel.attention_cp.get_forward_context'
+        'vllm_ascend.ascend_forward_context.get_forward_context'
     )
     @patch_distributed_groups(dcp_size=2, pcp_size=2)
     def test_forward_decode_pcp_dcp(self, mock_all2all, mock_dcp, mock_pcp,

tests/ut/attention/test_attention_v1.py

@@ -212,7 +212,7 @@ class TestAscendAttentionBackendImpl(TestBase):
 
     @patch('torch_npu._npu_reshape_and_cache')
     @patch('torch_npu.npu_fused_infer_attention_score')
-    @patch('vllm_ascend.attention.attention_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     def test_forward_fused_infer_attention(
             self, mock_get_forward_context,
             mock_npu_fused_infer_attention_score, mock_npu_reshape_and_cache):
@@ -248,7 +248,7 @@ class TestAscendAttentionBackendImpl(TestBase):
     @patch('vllm_ascend.attention.attention_v1.using_paged_attention')
     @patch('torch_npu._npu_paged_attention')
     @patch('torch_npu._npu_reshape_and_cache')
-    @patch('vllm_ascend.attention.attention_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     def test_forward_paged_attention(self, mock_get_forward_context,
                                      mock_npu_reshape_and_cache,
                                      mock_paged_attention,
@@ -279,7 +279,7 @@ class TestAscendAttentionBackendImpl(TestBase):
         mock_paged_attention.assert_called_once()
         assert output.shape == (4, 8 * 64)
 
-    @patch('vllm_ascend.attention.attention_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     @patch('torch_npu.npu_fused_infer_attention_score')
     @patch('torch_npu._npu_reshape_and_cache')
     def test_forward_decode_only_swa(self, mock_npu_reshape_and_cache,
@@ -311,7 +311,7 @@ class TestAscendAttentionBackendImpl(TestBase):
         mock_fused_infer_attention_score.assert_called_once()
         assert output.shape == (10, 8, 64)
 
-    @patch('vllm_ascend.attention.attention_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     @patch('torch_npu._npu_paged_attention')
     @patch('torch_npu.npu_fused_infer_attention_score')
     @patch('torch_npu._npu_reshape_and_cache')

tests/ut/attention/test_mla_cp.py

@@ -449,7 +449,7 @@ class TestAscendMLAImpl(TestBase):
         self.assertEqual(result.shape[1], N)
         self.assertEqual(result.shape[2], self.impl.kv_lora_rank + 1)
 
-    @patch('vllm_ascend.attention.context_parallel.mla_cp.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     @patch("torch_npu.npu_fused_infer_attention_score")
     @patch('torch_npu.npu_attention_update')
     @patch_distributed_groups(dcp_size=2, pcp_size=2, needs_mocks=False)

tests/ut/attention/test_mla_v1.py

@@ -929,7 +929,7 @@ class TestAscendMLAImpl(TestBase):
         self.assertEqual(out.shape, prefix_out.shape)
         self.assertEqual(lse.shape, prefix_lse.shape)
 
-    @patch('vllm_ascend.attention.mla_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     @patch("vllm_ascend.attention.mla_v1.AscendMLAImpl._v_up_proj")
     @patch("torch_npu.npu_fused_infer_attention_score")
     def test_forward_decode_without_graph(self,
@@ -1095,7 +1095,7 @@ class TestAscendMLAImpl(TestBase):
         self.assertEqual(k_pe.shape[-1], self.impl.qk_rope_head_dim)
         self.assertEqual(k_nope.shape[-1], self.impl.kv_lora_rank)
 
-    @patch('vllm_ascend.attention.mla_v1.get_forward_context')
+    @patch('vllm_ascend.ascend_forward_context.get_forward_context')
     @patch("torch_npu.npu_fused_infer_attention_score")
     def test_forward_decode(self, mock_npu_fused_infer_attention_score,
                             mock_get_forward_context):