[MISC] Cherry pick #1291 from v0.9.1-dev (#1825)

### What this PR does / why we need it?
Cherry pick #1291 from v0.9.1-dev, This pr implement the synchronization
of whether `dbo` is enabled across all dp ranks. specifically, it
performed allreduce op across multiple DP ranks, only when all the dp
rank is `enable_dbo`, it is enabled

Co-authored-by: shikang-hangzhou <459956190@qq.com>
Co-authored-by: wangli <wangli858794774@gmail.com>

- vLLM version: v0.10.0
- vLLM main:
2836dd73f1

---------

Signed-off-by: wangli <wangli858794774@gmail.com>

vllm_ascend/attention/mla_v1.py

@@ -137,6 +137,7 @@ class AscendMLAMetadata:
 
     decode: Optional[AscendMLADecodeMetadata] = None
     prefill: Optional[AscendMLAPrefillMetadata] = None
+    enable_dbo_across_dp: bool = False
 
     def __post_init__(self):
         pass
@@ -370,6 +371,7 @@ class AscendMLAMetadataBuilder:
         max_query_len: int,
         graph_pad_size: int = -1,
         query_start_loc: torch.Tensor = None,
+        enable_dbo_across_dp: bool = False,
     ) -> AscendMLAMetadata:
         assert self._num_decodes + self._num_prefills == num_reqs
 
@@ -536,6 +538,7 @@ class AscendMLAMetadataBuilder:
             query_start_loc=query_start_loc,
             block_tables=block_table,
             seq_lens=seq_lens,
+            enable_dbo_across_dp=enable_dbo_across_dp,
         )