[NPU]: Add NPU support for the swiglu

[jiaqiw09]

Summary

Add NPU support for the swiglu.

Details

• Implements a flattened, grid-stride Triton kernel for SwiGLU forward/backward to improve scalability and reduce launch overhead on Ascend NPUs.
• Uses UB-aware tiling (compute_default_tiling_strategy) and NPU vector core count to dynamically select block size and grid size for better performance stability.

Testing Done

I tested swiglu by following method and all cases passed:

• python benchmark/scripts/benchmark_swiglu.py

• pytest -v test/transformers/test_swiglu.py

• run make test to ensure correctness

• run make checkstyle to ensure code style

• run make test-convergence to ensure convergence

[jiaqiw09]

@Tcc0403 would you mind having a preview?

[jiaqiw09]

@Tcc0403 thx for you review, I have just updated the code.

[Tcc0403]

LGTM