[Model] Support pooling models (#3122)

### What this PR does / why we need it?

Support pooling models (like `bge-reranker-v2-m3`) in vllm-ascend, this
pr covered the three model types of embed (cls_token, mean_token,
lasttoken).

After this
[commit](17373dcd93),
vllm has provided support for adapting pooling models on the v1 engine.
This PR includes corresponding adaptations on the vllm-ascend side.

Fixes #1960

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

---------

Signed-off-by: lianyibo <lianyibo1@kunlunit.com>
Signed-off-by: MengqingCao <cmq0113@163.com>
Co-authored-by: MengqingCao <cmq0113@163.com>

vllm_ascend/worker/npu_input_batch.py

@@ -793,17 +793,12 @@ class InputBatch:
             logitsprocs=self.logitsprocs,
         )
 
-    @property
-    def pooling_metadata(self) -> PoolingMetadata:
-        if len(self.pooling_params) == 0:
-            pooling_params = []
-        else:
-            # Note, for now this assumes that all request in the batch
-            # are either sampling or pooling requests
-            assert len(self.req_ids) == len(self.pooling_params)
-            pooling_params = [
-                self.pooling_params[req_id] for req_id in self.req_ids
-            ]
+    def get_pooling_params(self) -> list[PoolingParams]:
+        assert len(self.req_ids) == len(self.pooling_params)
+        return [self.pooling_params[req_id] for req_id in self.req_ids]
+
+    def get_pooling_metadata(self) -> PoolingMetadata:
+        pooling_params = self.get_pooling_params()
 
         return PoolingMetadata(
             prompt_lens=torch.from_numpy(