[Bugfix] fix mtp profile run error where main model and mtp model use different quantization (#4102)

### What this PR does / why we need it?
In PR https://github.com/vllm-project/vllm-ascend/pull/3420, we
initially placed the quantization type (quant_type) in the MoECommMethod
class. However, since MoECommMethod follows a singleton pattern, it
couldn't accommodate scenarios where different layers in the model might
use different quantization approaches (e.g., MTP modules using
floating-point computation while the main model employs quantized
computation).
In this PR, we've moved the quantization type to the AscendFusedMoe
class and pass it as a parameter to MoECommMethod.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
```bash
export HCCL_BUFFSIZE=1024
export VLLM_VERSION=0.11.0

vllm serve /home/data/DeepSeek-R1_w8a8/ \
 --data-parallel-size 2 \
 --tensor-parallel-size 8 \
 --enable-expert-parallel \
 --served-model-name dsv3 \
 --max-model-len 32768 \
 --max-num-batched-tokens 4096 \
 --max-num-seqs 16 \
 --quantization ascend \
 --trust-remote-code \
 --gpu-memory-utilization 0.9 \
 --speculative-config '{"num_speculative_tokens": 2, "method":"deepseek_mtp"}'
```


- vLLM version: v0.11.0
- vLLM main:
83f478bb19

---------

Signed-off-by: realliujiaxu <realliujiaxu@163.com>

vllm_ascend/ops/fused_moe/fused_moe.py

@@ -37,6 +37,11 @@ from vllm_ascend.eplb.core.eplb_utils import (determine_default_expert_map,
 from vllm_ascend.ops.expert_load_balancer import ExpertLoadBalancer
 from vllm_ascend.ops.fused_moe.experts_selector import select_experts
 from vllm_ascend.ops.fused_moe.moe_comm_method import setup_moe_comm_method
+from vllm_ascend.ops.fused_moe.prepare_finalize import QuantType
+from vllm_ascend.quantization.w4a8_dynamic import \
+    AscendW4A8DynamicFusedMoEMethod
+from vllm_ascend.quantization.w8a8_dynamic import \
+    AscendW8A8DynamicFusedMoEMethod
 from vllm_ascend.utils import (ACL_FORMAT_FRACTAL_NZ, enable_sp, is_310p,
                                is_enable_nz, npu_stream_switch,
                                shared_expert_dp_enabled,
@@ -289,7 +294,23 @@ class AscendFusedMoE(FusedMoE):
 
         self.enable_shared_expert_dp = ascend_config.enable_shared_expert_dp
 
-        setup_moe_comm_method(self.moe_config, self.quant_method)
+        setup_moe_comm_method(self.moe_config)
+        self.quant_type = self._get_quant_type()
+
+    def _get_quant_type(self) -> QuantType:
+        quant_method = self.quant_method
+        if not hasattr(quant_method,
+                       "quant_method") or quant_method.quant_method is None:
+            return QuantType.NONE
+
+        method = quant_method.quant_method
+
+        if isinstance(method, AscendW8A8DynamicFusedMoEMethod):
+            return QuantType.W8A8
+        elif isinstance(method, AscendW4A8DynamicFusedMoEMethod):
+            return QuantType.W4A8
+        else:
+            return QuantType.NONE
 
     def update_expert_map(self, new_expert_map):
         self.expert_map = new_expert_map
@@ -334,7 +355,8 @@ class AscendFusedMoE(FusedMoE):
             hidden_states=hidden_states,
             router_logits=router_logits,
             replace_allreduce=forward_context.sp_enabled,
-            enable_shared_expert_dp=self.enable_shared_expert_dp)
+            enable_shared_expert_dp=self.enable_shared_expert_dp,
+            quant_type=self.quant_type)
 
         if isinstance(hidden_states, tuple):
             hidden_states, pertoken_scale = hidden_states