[MM][Perf] Enable 2.7x faster for convolution computation with aclnn BatchMatMulV2 (#7017)

### What this PR does / why we need it?
Currently, we are using
e2b31243c0/vllm/model_executor/layers/conv.py (L219-L232)
for convolution computation, which is used in patch embedding for VL
models.

After profiling, we find that this linear method will take about **6.87
ms**, which is much slower than just using `F.conv3d()`. In
`F.conv3d()`, it will call aclnn `BatchMatMulV2` with optimization on
Ascend NPU, which only take about **2.50 ms** and is **2.7x faster**
than linear method.

- vLLM version: v0.16.0
- vLLM main:
15d76f74e2
---------
Signed-off-by: shen-shanshan <467638484@qq.com>

vllm_ascend/utils.py

@@ -597,6 +597,7 @@ def register_ascend_customop(vllm_config: VllmConfig | None = None):
     from vllm.model_executor.custom_op import CustomOp
 
     from vllm_ascend.ops.activation import AscendQuickGELU, AscendSiluAndMul
+    from vllm_ascend.ops.conv import AscendConv2dLayer, AscendConv3dLayer
     from vllm_ascend.ops.fused_moe.fused_moe import AscendFusedMoE, AscendSharedFusedMoE
     from vllm_ascend.ops.layernorm import AscendGemmaRMSNorm, AscendRMSNorm, AscendRMSNormGated
     from vllm_ascend.ops.linear import (
@@ -645,6 +646,8 @@ def register_ascend_customop(vllm_config: VllmConfig | None = None):
         "MMEncoderAttention": AscendMMEncoderAttention,
         "ApplyRotaryEmb": AscendApplyRotaryEmb,
         "RMSNormGated": AscendRMSNormGated,
+        "Conv2dLayer": AscendConv2dLayer,
+        "Conv3dLayer": AscendConv3dLayer,
     }
 
     # 310P: override selected ops with 310P implementations (keep minimal changes outside _310p)