[refact] unified soc_version code (#4359)

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

Currently, there are two paths to judge the chip type in code,
`get_ascend_soc_version` use `get_soc_version` api in torch_npu, and
`is_310p` `use _build_info.__soc_version__`, which generate when
install. We need to unify the two paths.

We need to unify these codes based on the following points:

1. We need to ensure consistency in chip type judgment between compiling
and running states;
2. In compiling state, we need chip type to complete op's compilation,
but in running state, we only need device
type(910B/910_93/310P/910_95/etc) to make code branch judgement;
3. In compiling state, torch_npu may not have been installed yet, so we
can't use torch_npu's api.

Based on the above points, we have made the following changes:

1. When user set env `SOC_VERSION`, use it; when not set, query
soc_version by `npu-smi`;
2. generate device_type based on soc_version when compiling, and write
`__device_type__` instead of `__soc_version__` in `_build_info.py`;
3. In running state, use `__device_type__` to judge code branch.

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

When not set env `SOC_VERSION`, it will not be `ASCEND910B1` by default,
we will query soc_version by `npu-smi`. And env `SOC_VERSION` must be in
the list `soc_to_device` in `setup.py`.

- vLLM version: v0.11.0
- vLLM main:
2918c1b49c

Signed-off-by: zzzzwwjj <1183291235@qq.com>

tests/ut/quantization/test_w8a8.py

@@ -12,6 +12,7 @@ from vllm_ascend.quantization.w8a8 import (AscendC8KVCacheMethod,
                                            AscendW8A8LinearMethod,
                                            fused_experts, fused_experts_310p,
                                            quant_per_tensor)
+from vllm_ascend.utils import AscendDeviceType
 
 
 class TestQuantPerTensor(TestBase):
@@ -118,9 +119,11 @@ class TestAscendW8A8LinearMethod(TestBase):
         expected_y_output += bias
         self.assertTrue(torch.equal(output, expected_y_output))
 
-    @patch("vllm_ascend.quantization.w8a8.is_310p", return_value=True)
+    @patch('vllm_ascend.utils.get_ascend_device_type',
+           return_value=AscendDeviceType._310P)
     @patch("torch_npu.npu_quant_matmul")
-    def test_apply_with_x_is_310p(self, mock_npu_quant_matmul, mock_is_310p):
+    def test_apply_with_x_is_310p(self, mock_npu_quant_matmul,
+                                  mock_soc_version):
         layer = MagicMock()
         layer.aclnn_input_scale = 0.1
         layer.aclnn_input_offset = 0.2
@@ -279,11 +282,12 @@ class TestAscendW8A8FusedMoEMethod(TestBase):
         mock_fused_experts.assert_called_once()
         self.assertEqual(result.shape, (32, self.hidden_size))
 
-    @patch("vllm_ascend.quantization.w8a8.is_310p", return_value=True)
+    @patch('vllm_ascend.quantization.w8a8.get_ascend_device_type',
+           return_value=AscendDeviceType._310P)
     @patch('vllm_ascend.quantization.w8a8.select_experts')
     @patch('vllm_ascend.quantization.w8a8.fused_experts_310p')
     def test_apply_is_310p(self, mock_fused_experts_310p, mock_select_experts,
-                           mock_is_310p):
+                           mock_soc_version):
         # Setup
         mock_layer = MagicMock()
         x = torch.randn(32, self.hidden_size)
@@ -342,8 +346,9 @@ class TestAscendC8KVCacheMethod(TestBase):
             expected_shape = (self.layer.num_kv_heads * self.layer.head_size, )
             self.assertEqual(param.shape, expected_shape)
 
-    @patch("vllm_ascend.quantization.w8a8.is_310p", return_value=False)
-    def test_process_weights_after_loading_not_310p(self, mock_is_310p):
+    @patch('vllm_ascend.utils.get_ascend_device_type',
+           return_value=AscendDeviceType._910_93)
+    def test_process_weights_after_loading_not_310p(self, mock_soc_version):
         key_data = torch.ones(4 * 64)
         value_data = torch.ones(4 * 64) * 2
 
@@ -356,8 +361,9 @@ class TestAscendC8KVCacheMethod(TestBase):
         self.assertTrue(torch.all(self.method.antiquant_scale_comb[0] == 1))
         self.assertTrue(torch.all(self.method.antiquant_scale_comb[1] == 2))
 
-    @patch("vllm_ascend.quantization.w8a8.is_310p", return_value=True)
-    def test_process_weights_after_loading_is_310p(self, mock_is_310p):
+    @patch('vllm_ascend.utils.get_ascend_device_type',
+           return_value=AscendDeviceType._310P)
+    def test_process_weights_after_loading_is_310p(self, mock_soc_version):
         key_data = torch.ones(4 * 64)
         value_data = torch.ones(4 * 64) * 2