feat(attention_cp): support chunked prefill for Qwen3Next with PCP&DCP (#6900)

### What this PR does / why we need it?
Support chunked prefill for Qwen3Next with PCP&DCP

- vLLM version: v0.16.0
- vLLM main:
15d76f74e2

---------

Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>

vllm_ascend/attention/utils.py

@@ -113,6 +113,10 @@ class AscendPrefillContextParallelMetadata:
     # when entering from linear-attention to attention
     pcp_enter_fa_restore_idx: torch.Tensor = None
 
+    # scatter the full sequence across all pcp ranks
+    # when exiting from attention to linear-attention
+    pcp_exit_fa_scatter_idx: torch.Tensor = None
+
     # the number of tokens padded in linear-attn per rank
     pcp_padded_tokens_fla: int = 0