support async mtp (#4511)

### What this PR does / why we need it?
this pr aims to support async_scheduling for mtp, which refer to vllm pr
https://github.com/vllm-project/vllm/pull/24799.
and this pr fix some synchronize problem in vllm-ascend.
### Does this PR introduce _any_ user-facing change?

### How was this patch tested?


- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: Ronald1995 <ronaldautomobile@163.com>
Co-authored-by: wangxiyuan <wangxiyuan1007@gmail.com>

vllm_ascend/worker/npu_input_batch.py

@@ -68,6 +68,8 @@ class CachedRequestState:
     lora_request: Optional[LoRARequest] = None
     prompt_embeds: Optional[torch.Tensor] = None
 
+    prev_num_draft_len: int = 0  # previous number of draft tokens
+
     def __post_init__(self):
         self.num_prompt_tokens = length_from_prompt_token_ids_or_embeds(
             self.prompt_token_ids, self.prompt_embeds)