[Perf] Avoid performing index selection of sin/cos cache every layer (#1890)

Optimize number of index selections of sin/cos cache.

- vLLM version: v0.10.0
- vLLM main:
656c24f1b5

Signed-off-by: whx-sjtu <2952154980@qq.com>

vllm_ascend/worker/model_runner_v1.py

@@ -1799,6 +1799,9 @@ class NPUModelRunner(LoRAModelRunnerMixin):
                             attn_metadata.decode.input_positions)
                         torch._dynamo.mark_static(
                             get_forward_context().mc2_mask)
+                        if hasattr(attn_metadata.decode, "sin"):
+                            torch._dynamo.mark_static(attn_metadata.decode.sin)
+                            torch._dynamo.mark_static(attn_metadata.decode.cos)
                         torch._dynamo.mark_static(attn_metadata.slot_mapping)
                         for kv in self.kv_caches:
                             assert isinstance(