feat(ascend): basic llama operators

[wooway777]

Summary

• Add Ascend ACLNN implementations for Argmax, CausalSoftmax, Copy, FusedAddRmsNorm, RmsNorm, RotaryEmbedding, and SiluAndMul.
• Cover the core operator paths required by Llama/Qwen inference, including BF16 and mixed-dtype normalization, non-contiguous tensors, repeated execution, and NeoX-style rotary embedding.

Motivation

Complete the Ascend operator coverage required by the InfiniLM inference path for Llama/Qwen-family models. The implementations use CANN ACLNN operators while preserving the semantics and numerical precision of the verified InfiniCore implementations.

Closes # N/A

Type of Change

• feat — new feature / new operator / new platform
• fix — bug fix
• perf — performance improvement (no behavioral change)
• refactor — code restructuring without behavior change
• test — adding or fixing tests only
• docs — documentation only
• build / ci — build system or CI configuration
• chore — tooling, formatting, or other non-code changes
• Breaking change (requires a ! in the Conventional Commits prefix or a BREAKING CHANGE: footer)

Platforms Affected

• CPU (WITH_CPU)
• NVIDIA (WITH_NVIDIA)
• Iluvatar (WITH_ILUVATAR)
• MetaX (WITH_METAX)
• Cambricon (WITH_CAMBRICON)
• Moore (WITH_MOORE)
• Ascend (WITH_ASCEND)
• PyTorch C++ bindings (WITH_TORCH)
• Build system / CMake / CI
• Python bindings / user-facing API

Smoke Test Result

paste smoke test output here

Test Results on Supported Platforms

Platform	Affected	Build / Smoke Result	Full Result / Notes
NVIDIA
Iluvatar
MetaX
Cambricon
Moore
Ascend

Full `pytest` output (optional)

paste here

Benchmark / Performance Impact

Notes for Reviewers