[Main2Main] Upgrade vllm commit to 0105 (#5595)

### What this PR does / why we need it?

Upgrade vllm commit to 0105 (8be6432bdaf6275664d857b1e5e9bf8ed1ce299e)

1. Remove `maybe_padded_num_tokens` arg in `model_runner_v1.py` since
https://github.com/vllm-project/vllm/pull/31517 deleted unused arg

2. Remove dense `Qwen/Qwen3-0.6B` in
`tests/e2e/multicard/test_aclgraph_capture_replay.py` and
`tests/e2e/multicard/test_data_parallel.py` due to
https://github.com/vllm-project/vllm/pull/30739
where offline data parallel mode will not be supported/useful for dense
models

3. Adapt `vllm_ascend/worker/worker.py` due to
https://github.com/vllm-project/vllm/pull/31584

4. Adapt `self.block_size` calling due to
https://github.com/vllm-project/vllm/pull/31540

5. Modify `test_mla_v1.py` due to
https://github.com/vllm-project/vllm/pull/28454 , which refactorred
`get_head_size()`

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.13.0
- vLLM main:
7157596103

Signed-off-by: wjunLu <wjunlu217@gmail.com>

tests/e2e/multicard/test_aclgraph_capture_replay.py

@@ -28,7 +28,8 @@ from vllm.utils.network_utils import get_open_port
 from vllm_ascend.utils import AscendDeviceType, get_ascend_device_type
 
 MODELS = [
-    "Qwen/Qwen3-0.6B",
+    # Offline data parallel mode will be not supported/useful for dense models
+    # "Qwen/Qwen3-0.6B",
     "vllm-ascend/DeepSeek-V2-Lite-W8A8",
 ]
 

tests/e2e/multicard/test_data_parallel.py

@@ -27,9 +27,7 @@ from unittest.mock import patch
 
 import pytest
 
-MODELS = [
-    "Qwen/Qwen3-0.6B", "Qwen/Qwen3-30B-A3B", "vllm-ascend/Qwen3-30B-A3B-W8A8"
-]
+MODELS = ["Qwen/Qwen3-30B-A3B", "vllm-ascend/Qwen3-30B-A3B-W8A8"]
 
 
 @pytest.mark.parametrize("model", MODELS)

tests/e2e/multicard/test_data_parallel_tp2.py

@@ -9,7 +9,7 @@ from unittest.mock import patch
 
 import pytest
 
-MODELS = ["Qwen/Qwen3-0.6B"]
+MODELS = ["Qwen/Qwen3-30B-A3B"]
 
 
 @pytest.mark.parametrize("model", MODELS)

tests/ut/attention/test_mla_v1.py

@@ -17,6 +17,7 @@ from vllm_ascend.attention.mla_v1 import (AscendMLABackend,
                                           AscendMLAPrefillMetadata,
                                           ChunkedContextMetadata)
 from vllm_ascend.attention.utils import AscendCommonAttentionMetadata
+from vllm_ascend.utils import vllm_version_is
 
 
 class TestAscendMLABackend(TestBase):
@@ -392,7 +393,10 @@ class TestAscendMLAMetadataBuilderBuild(TestBase):
         self.mock_vllm_config.model_config = model_config
         self.kv_cache_spec = MagicMock()
         self.kv_cache_spec.num_layers = 32
-        self.kv_cache_spec.head_size = 128
+        if vllm_version_is('0.13.0'):
+            self.kv_cache_spec.head_size = 128
+        else:
+            self.kv_cache_spec.head_size = 64
         self.kv_cache_spec.num_heads = 32
 
     @patch("vllm_ascend.attention.mla_v1.get_cos_and_sin_mla")

tests/ut/conftest.py

@@ -18,13 +18,6 @@
 import sys
 from unittest.mock import MagicMock
 
-from vllm_ascend.utils import adapt_patch  # noqa E402
-from vllm_ascend.utils import register_ascend_customop
-
-# triton and torch_npu is not available in the environment, so we need to mock them
-sys.modules['torch_npu'].npu.current_device = MagicMock(return_value=0)
-sys.modules['torch_npu._inductor'] = MagicMock()
-
 triton_runtime = MagicMock()
 triton_runtime.driver.active.utils.get_device_properties.return_value = {
     'num_aic': 8,
@@ -32,6 +25,13 @@ triton_runtime.driver.active.utils.get_device_properties.return_value = {
 }
 sys.modules['triton.runtime'] = triton_runtime
 
+from vllm_ascend.utils import adapt_patch  # noqa E402
+from vllm_ascend.utils import register_ascend_customop  # noqa E402
+
+# triton and torch_npu is not available in the environment, so we need to mock them
+sys.modules['torch_npu'].npu.current_device = MagicMock(return_value=0)
+sys.modules['torch_npu._inductor'] = MagicMock()
+
 adapt_patch()
 adapt_patch(True)
 

tests/ut/spec_decode/test_eagle_proposer.py

@@ -58,7 +58,6 @@ class TestEagleProposerInitialization(TestBase):
                                  device=self.device,
                                  runner=self.runner)
 
-        self.assertEqual(proposer.block_size, 16)
         self.assertEqual(proposer.hidden_size, 4096)
         self.assertTrue(proposer.use_cuda_graph)
 

tests/ut/spec_decode/test_mtp_proposer.py

@@ -86,7 +86,6 @@ class TestMtpProposer:
         assert proposer.dtype == torch.float16
         assert proposer.num_speculative_tokens == 2
         assert proposer.hidden_size == 4096
-        assert proposer.block_size == 16
 
         # Test with mrope enabled
         assert hasattr(proposer, "positions")