[BugFix]Fix precision issue for LoRA feature (#4141)

vLLM version: v0.11.0
vLLM main: vllm-project/vllm

### What this PR does / why we need it?
   Fix the precision issue of the LoRA feature in vllm-ascend.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?
```bash
pytest tests/lora/test_llama_tp.py::test_llama_lora -s
```
<img width="1319" height="879" alt="lora_test"
src="https://github.com/user-attachments/assets/2a0b2325-5b05-4bbc-ac03-a7c9f0ad9d4c"
/>


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

---------

Signed-off-by: hukongyi <hukongyi@cmbchina.com>

csrc/kernels/sgmv_shrink.cpp

@@ -242,7 +242,7 @@ private:
 
 // declare all dtype kernel
 SGMV_SHRINK_TYPE_DECLARE(half)
-#if (__CCE_AICORE__ >= 220)
+#if !defined(__CCE_AICORE__) || (__CCE_AICORE__ >= 220)
     SGMV_SHRINK_TYPE_DECLARE(bfloat16_t)
 #endif
 
@@ -260,8 +260,8 @@ extern void sgmv_shrink_impl(AscendType type, void* stream, void* x, void* weigh
                                                         numTokensPerCore, inputHiddenDim, maxLoRARank,
                                                         scale);
     } else if (type == AscendType::BF16) {
-        #if (__CCE_AICORE__ >= 220)
-            sgmv_shrink_bfloat16_t<<<blockDim, nullptr, stream>>>(x, weight, loraIndices, loraIndicesSize, 
+        #if !defined(__CCE_AICORE__) || (__CCE_AICORE__ >= 220)
+        sgmv_shrink_bfloat16_t<<<blockDim, nullptr, stream>>>(x, weight, loraIndices, loraIndicesSize, 
                                                                   seqLen, seqLenSize, 
                                                                   y, batchSize,
                                                                   numTokensPerCore, inputHiddenDim, maxLoRARank,