[feat]: oproj tensor parallelism in pure DP and graph-mode scenarios. (#2167)

### What this PR does / why we need it?
This PR introduces Oproj matrix tensor model parallel to achieve
decreasing of memory consumption. It only support graph mode in pure DP
scenario.

In deepseek r1 w8a8 PD disagregated Decode instance, using pure DP, with
oproj_tensor_parallel_size = 8, we have 1 ms TPOT increasing, saved 5.8
GB NPU memory per RANK. We got best performance when
oproj_tensor_parallel_size=4 without TPOT increasing.

performance data:
<img width="1442" height="442" alt="image"
src="https://github.com/user-attachments/assets/83270fc5-868a-4387-b0a9-fac29b4a376d"
/>

### Does this PR introduce _any_ user-facing change?
This PR introduces one new config in `additional_config`.
| Name | Effect | Required | Type | Constraints |
| :---------------------------- |
:--------------------------------------- | :------- | :--- |
:----------------- |
| oproj_tensor_parallel_size | Split the o_proj matrix along the row
dimension (head num * head dim) into oproj_tensor_parallel_size pieces.
| No | int | default value is None, once this value is set, the feature
will be enabled, head num * head dim must be divisible by this value. |

example

`--additional_config={"oproj_tensor_parallel_size": 8}`

### How was this patch tested?


- vLLM version: v0.10.1.1
- vLLM main:
eddaafc1c7

---------

Signed-off-by: zzhx1 <zzh_201018@outlook.com>
Co-authored-by: zzh <zzh_201018@outlook.com>

vllm_ascend/quantization/quant_config.py

@@ -35,8 +35,11 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
 from vllm.model_executor.parameter import PerTensorScaleParameter
 from vllm.model_executor.utils import set_weight_attrs
 
+from vllm_ascend.distributed.parallel_state import (get_mlp_tp_group,
+                                                    get_otp_group)
 from vllm_ascend.ops.fused_moe import AscendUnquantizedFusedMoEMethod
-from vllm_ascend.utils import ASCEND_QUANTIZATION_METHOD
+from vllm_ascend.utils import (ASCEND_QUANTIZATION_METHOD, mlp_tp_enable,
+                               oproj_tp_enable)
 
 from .utils import get_quant_method
 
@@ -220,9 +223,15 @@ class AscendLinearMethod(LinearMethodBase):
         bias: Optional[torch.Tensor] = None,
     ) -> torch.Tensor:
         if isinstance(layer, RowParallelLinear):
-            tp_rank = get_tensor_model_parallel_rank()
-            return self.quant_method.apply(layer, x, bias, tp_rank)
-        return self.quant_method.apply(layer, x, bias)
+            if layer.prefix.find("o_proj") != -1 and oproj_tp_enable():
+                tp_rank = get_otp_group().rank_in_group
+            elif layer.prefix.find("down_proj") != -1 and mlp_tp_enable():
+                tp_rank = get_mlp_tp_group().rank_in_group
+            else:
+                tp_rank = get_tensor_model_parallel_rank()
+        else:
+            tp_rank = 0
+        return self.quant_method.apply(layer, x, bias, tp_rank)
 
 
 class AscendKVCacheMethod(BaseKVCacheMethod):