[310P]: refactoring for 310p kvcache and some ops class (#6117)

### What this PR does / why we need it?
* Refactor the LayerNorm and activation operator classes to decouple the
310P device implementation from the main branch.
* Refactor `mm_encoder_attention` on 310P to use the
`torch_npu._npu_flash_attention_unpad` operator.
* Refactor the QKV inputs in the prefill stage of `attention_v1` on 310P
so they are no longer padded to 16× alignment.
* Refactor `model_runner` on 310P to align the KV-cache initialization
logic with the mainline implementation.

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

### How was this patch tested?
use the e2e tests.

- vLLM version: v0.13.0
- vLLM main:
d68209402d

---------

Signed-off-by: Tflowers-0129 <2906339855@qq.com>

.github/workflows/schedule_image_build_and_push.yaml

@@ -43,10 +43,10 @@ jobs:
           - name: A3 openEuler
             dockerfile: Dockerfile.a3.openEuler
             suffix: 'a3-openeuler'
-          # - name: 310P Ubuntu
-          #   dockerfile: Dockerfile.310p
-          # - name: 310P openEuler
-          #   dockerfile: Dockerfile.310p.openEuler
+          - name: 310P Ubuntu
+            dockerfile: Dockerfile.310p
+          - name: 310P openEuler
+            dockerfile: Dockerfile.310p.openEuler
     uses: ./.github/workflows/_schedule_image_build.yaml
     with:
       dockerfile: ${{ matrix.build_meta.dockerfile }}