[torchair]remove aicpu op (vllm-project#2640)

### What this PR does / why we need it?
remove aicpu op for torchair mode
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
vLLM version: v0.10.1.1
vLLM main:
vllm-project/vllm@05d839c
- vLLM version: v0.10.1.1
- vLLM main:
vllm-project/vllm@67c1490

Signed-off-by: zhangdepeng <zhangdepeng2@huawei.com>
Co-authored-by: zhangdepeng <zhangdepeng2@huawei.com>

Author: 2 people

vllm_ascend/torchair/torchair_attention.py

@@ -304,6 +304,7 @@ def __init__(
         self.num_queries_per_kv = self.num_heads // self.num_kv_heads
         self.key_cache = None
         self.value_cache = None
+        self.scale_tensor = torch.zeros((), device='npu', dtype=torch.int32)
 
     def forward(
         self,
@@ -366,7 +367,7 @@ def forward(
             key_cache, value_cache = kv_cache[0], kv_cache[1]
             slots = attn_metadata.slot_mapping
 
-            block_size = key_cache.shape[1]
+            block_size = self.scale_tensor + key_cache.shape[1]
             slots_indices = slots.reshape(-1, 1)
             block_indices = slots_indices // block_size
             slots_indices = slots_indices % block_size