shared_experts+router_experts merge all_reduce(Improve TTOP 5ms) (#1395)

### What this PR does / why we need it?
When all_reduce_merge is in progress, shared_experts does not do
all_reduce in mlp, but waits until shared_experts+router_experts are
completed before doing all_reduce
In prefill and decode, as long as shared_experts+router_experts are
all_reduce, there will be benefits.
### Does this PR introduce _any_ user-facing change?

### How was this patch tested?
bash examples/run_dp_attention_etp16.sh
bash examples/run_dp_attention_etp16_benmark.sh
- vLLM version: v0.9.1
- vLLM main:
977180c912

---------

Signed-off-by: ttanzhiqiang <389825161@qq.com>

examples/run_dp_attention_etp16.sh

@@ -3,9 +3,10 @@ export TASK_QUEUE_ENABLE=1
 source /usr/local/Ascend/ascend-toolkit/set_env.sh
 source /usr/local/Ascend/nnal/atb/set_env.sh
 export ASCEND_LAUNCH_BLOCKING=0
-export VLLM_VERSION=0.9.0
+export VLLM_VERSION=0.9.1
 
 nohup python -m vllm.entrypoints.openai.api_server --model=/mnt/deepseek/DeepSeek-R1-W8A8-VLLM \
+    --served-model-name auto \
     --quantization ascend \
     --trust-remote-code \
     --distributed-executor-backend=mp \

examples/run_dp_attention_etp16_benmark.sh

@@ -21,7 +21,8 @@ for concurrency in "${concurrency_array[@]}"; do
         python /mnt/deepseek/vllm/benchmarks/benchmark_serving.py \
             --backend vllm \
             --trust-remote-code \
-            --model /mnt/deepseek/DeepSeek-R1-W8A8-VLLM \
+            --model auto \
+            --tokenizer /mnt/deepseek/DeepSeek-R1-W8A8-VLLM \ 
             --dataset-name random \
             --random-input-len 4096 \
             --random-output-len 1536 \