fix: use InfiniRT Hygon runtime headers

[voltjia]

Summary

• Replace the local Hygon Runtime and DeviceEnabled definitions with the corresponding InfiniRT Hygon headers.
• Add Hygon data_type_.h and caster.cuh wrappers so InfiniOps keeps its platform caster specialization while data type ownership comes from InfiniRT.
• Include the Hygon InfiniRT device marker in generated dispatch/call-instantiation sources.

Motivation

Hygon still carried local runtime/device/type definitions after the InfiniRT runtime-header migration, while the other CUDA-like backends consume InfiniRT-owned headers. This keeps Hygon aligned with the current runtime ownership model and prevents generated Hygon dispatch code from missing the InfiniRT Hygon device marker.

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)
• Hygon (WITH_HYGON)
• Ascend (WITH_ASCEND)
• PyTorch C++ bindings (WITH_TORCH)
• Build system / CMake / CI
• Python bindings / user-facing API

Smoke Test Result

Hygon smoke was not run locally because this Windows workspace does not have cmake in PATH and does not have a Hygon/DTK device toolchain available. The repository Hygon CI job is the intended affected-platform smoke path for this PR; it builds InfiniRT with WITH_HYGON=ON, installs it, builds InfiniOps with INFINI_OPS_SMOKE_BUILD=ON, then runs pytest tests/ -m smoke.

Local checks run:

> git diff --check
passed

> C:\Users\huang\.cache\codex-runtimes\codex-primary-runtime\dependencies\python\python.exe -m py_compile scripts\generate_wrappers.py
passed

> C:\Users\huang\.cache\codex-runtimes\codex-primary-runtime\dependencies\python\python.exe -c "import importlib.util; spec=importlib.util.spec_from_file_location('gw','scripts/generate_wrappers.py'); m=importlib.util.module_from_spec(spec); spec.loader.exec_module(m); assert 'infini/rt/hygon/device_.h' in m._device_marker_headers(['hygon']); print('hygon marker ok')"
hygon marker ok

> rg -n "struct Runtime<Device::Type::kHygon>|struct DeviceEnabled<Device::Type::kHygon>|struct TypeMap<Device::Type::kHygon" -S src\native\cuda\hygon scripts\generate_wrappers.py
no matches

Test Results on Supported Platforms

Platform	Affected	Build / Smoke Result	Full Result / Notes
NVIDIA	No	N/A - not affected	N/A
Iluvatar	No	N/A - not affected	N/A
MetaX	No	N/A - not affected	N/A
Cambricon	No	N/A - not affected	N/A
Moore	No	N/A - not affected	N/A
Hygon	Yes	Not run locally - requires Hygon/DTK runner	Hygon CI smoke expected
Ascend	No	N/A - not affected	N/A

Full `pytest` output (optional)

N/A - not run locally; no Hygon/DTK environment in this workspace.

Benchmark / Performance Impact

N/A - this changes header ownership and generated include coverage only.

Notes for Reviewers

Please focus on the Hygon include chain: runtime_.h now consumes <infini/rt/hygon/runtime_.h> and keeps the InfiniOps-owned Caster<Device::Type::kHygon> specialization in native/cuda/hygon/caster.cuh so existing CUDA-like kernels still see the caster specialization.