[Triton][Config] Add muls_add triton kernel and refactor AscendCompilationConfig (#5518)

### What this PR does / why we need it?
Add muls_add triton kernel with related fusion pass. What's more, this
PR refactors `AscendCompilationConfig` and delete `NpugraphExConfig`.

### Does this PR introduce _any_ user-facing change?
None

### How was this patch tested?
CI passed with new added test.


- vLLM version: v0.13.0
- vLLM main:
45c1ca1ca1

---------

Signed-off-by: whx-sjtu <2952154980@qq.com>

tests/e2e/multicard/2-cards/test_sp_pass.py

@@ -28,7 +28,7 @@ def test_qwen3_vl_sp_tp2(model: str) -> None:
                 "cudagraph_mode": "FULL_DECODE_ONLY",
                 "pass_config": {"enable_sp": False}
             },
-            additional_config={"npugraph_ex_config": {"enable": False}}
+            additional_config={"ascend_compilation_config": {"enable_npugraph_ex": False}}
     ) as runner:
         no_sp_outputs = runner.model.generate(prompts, sampling_params)
 
@@ -41,7 +41,7 @@ def test_qwen3_vl_sp_tp2(model: str) -> None:
                 "cudagraph_mode": "FULL_DECODE_ONLY",
                 "pass_config": {"enable_sp": True}
             },
-            additional_config={"sp_threshold": 10, "npugraph_ex_config": {"enable": False}}
+            additional_config={"sp_threshold": 10, "ascend_compilation_config": {"enable_npugraph_ex": False}}
     ) as runner:
         sp_outputs = runner.model.generate(
             prompts, sampling_params)