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/attention/attention_v1.py

@@ -317,8 +317,8 @@ class AscendAttentionMetadataBuilder:
                     query_start_loc_cpu.device).to(query_start_loc_cpu.dtype)
             ])
 
-        query_start_loc = query_start_loc_cpu.to(self.device,
-                                                 non_blocking=True)
+        query_start_loc = query_start_loc_cpu.pin_memory().to(
+            self.device, non_blocking=True)
 
         attn_metadata = AscendMetadata(
             num_actual_tokens=num_actual_tokens,