[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>

vllm_ascend/torchair/models/torchair_pangu_moe.py

@@ -57,7 +57,8 @@ from vllm.sequence import IntermediateTensors
 from vllm.v1.sample.sampler import Sampler
 
 from vllm_ascend.ascend_config import get_ascend_config
-from vllm_ascend.utils import ACL_FORMAT_FRACTAL_NZ, is_310p
+from vllm_ascend.utils import (ACL_FORMAT_FRACTAL_NZ, AscendDeviceType,
+                               get_ascend_device_type)
 
 _ROUTER_SCALE = None
 
@@ -448,7 +449,8 @@ class PanguProMoESparseMoeBlock(nn.Module):
         # on 300I Duo platform, we find that num_voted_experts set to 5 achieves
         # good performance without sacrifice too much accuracy. for other platform,
         # this is set to 8 to use original pangu grouped topk.
-        num_voted_experts = 5 if is_310p() else 8
+        num_voted_experts = 5 if get_ascend_device_type(
+        ) == AscendDeviceType._310P else 8
 
         self.experts = FusedMoE(
             num_experts=config.num_experts,
@@ -1109,7 +1111,8 @@ class PanguProMoEForCausalLM(nn.Module, SupportsPP):
                                             default_weight_loader)
                     weight_loader(param, loaded_weight)
             loaded_params.add(name)
-            if is_310p() and "head" in name:
+            if get_ascend_device_type(
+            ) == AscendDeviceType._310P and "head" in name:
                 # on 300I Duo platform, ACL_FORMAT_FRACTAL_NZ is much more preferred than
                 # ACL_FORMAT_FRACTAL_ND by matmul operation. Since lmhead is also implemented
                 # by linear, we manually cast the format here.