[Feature] model_runner refactor (#4764)

### What this PR does / why we need it?
refactor npu_modelrunner， we should be close to gpu_modelrunner 

### Does this PR introduce _any_ user-facing change?
NO

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

---------

Signed-off-by: zhenwenqi2024 <zhenwenqi_2022@qq.com>
Signed-off-by: zhenwenqi2024 <155598497+zhenwenqi2024@users.noreply.github.com>

vllm_ascend/attention/utils.py

@@ -100,10 +100,6 @@ class AscendCommonAttentionMetadata:
     # padding tokens. It is used to handle some padding operations.
     num_input_tokens: int = 0
 
-    # NOTE: This is a temporary solution for rotary embedding in MLA
-    cos: torch.Tensor = None
-    sin: torch.Tensor = None
-
     prefill_context_parallel_metadata: Optional[
         AscendPrefillContextParallelMetadata] = None