[NPU] Add optimized NPU mhc (#1173)

Add Ascend NPU Triton kernels for the three mHC sub-operators:
  - Fused matmul + RMS normalization (forward/backward)
- Sinkhorn routing with split pre/post/residual coefficients
(forward/backward)
  - Pre-aggregate weighted sum (forward/backward)
  - Post + residual mixing (forward/backward)

  NPU optimizations applied:
  - Unified UB tiling via compute_default_tiling_strategy for matrix
  - Persistent grid-stride loops (tl.range + num_programs)
  - Adaptive BLOCK_N/BLOCK_M for core utilisation at small seq_len
  - Fused backward coefficient assembly kernel

Hardware Type: Atlas 800I A2
- [x] run `make test` to ensure correctness
- [x] run `make checkstyle` to ensure code style
- [ ] run `make test-convergence` to ensure convergence

Author: lowdy1

src/liger_kernel/ops/backends/_ascend/ops/__init__.py

@@ -55,6 +55,9 @@
 from liger_kernel.ops.backends._ascend.ops.llama4_rope import LigerLlama4RopeFunction
 from liger_kernel.ops.backends._ascend.ops.llama4_rope import llama4_rope_backward
 from liger_kernel.ops.backends._ascend.ops.llama4_rope import llama4_rope_forward
+from liger_kernel.ops.backends._ascend.ops.mhc import LigerMHCCoeffsFunction
+from liger_kernel.ops.backends._ascend.ops.mhc import LigerMHCPostResFunction
+from liger_kernel.ops.backends._ascend.ops.mhc import LigerMHCPreFunction
 from liger_kernel.ops.backends._ascend.ops.poly_norm import LigerPolyNormFunction
 from liger_kernel.ops.backends._ascend.ops.poly_norm import poly_norm_backward
 from liger_kernel.ops.backends._ascend.ops.poly_norm import poly_norm_forward
@@ -146,4 +149,7 @@
     "LigerFusedLinearCrossEntropyFunction",
     "fused_linear_cross_entropy_forward",
     "fused_linear_cross_entropy_backward",
+    "LigerMHCCoeffsFunction",
+    "LigerMHCPreFunction",
+    "LigerMHCPostResFunction",
 ]