feat(ascend): add fill provider

[baominghelly]

Summary

• Add the Ascend fill provider in src/native/ascend/ops/fill/kernel.h using the ACLNN in-place scalar and tensor fill APIs.
• Add Ascend device-scalar coverage to tests/test_fill.py while preserving the existing host-scalar cases.

Motivation

The generated fill interface did not have an Ascend slot-0 provider. This change enables both host scalar values and one-element device tensor values for model execution on Ascend.

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

bash /workspace/ascend-slot0/configure-focused-build.sh \
  /workspace/ascend-slot0/fill \
  /workspace/ascend-slot0/build-fill \
  /workspace/ascend-slot0/test-config/ops-fill.json
cmake --build /workspace/ascend-slot0/build-fill --parallel 32
bash /workspace/ascend-slot0/basic-ops-20260901/scripts/run-focused-test.sh \
  /workspace/ascend-slot0/fill \
  /workspace/ascend-slot0/build-fill \
  /workspace/ascend-slot0/python-fill \
  tests/test_fill.py

[100%] Built target ops
64 passed, 6 skipped, 15 warnings in 1.27s

Test Results on Supported Platforms

Platform	Affected	Build / Smoke Result	Full Result / Notes
NVIDIA	No	N/A - not affected	N/A - not affected
Iluvatar	No	N/A - not affected	N/A - not affected
MetaX	No	N/A - not affected	N/A - not affected
Cambricon	No	N/A - not affected	N/A - not affected
Moore	No	N/A - not affected	N/A - not affected
Ascend	Yes	Focused build passed on Ascend 910C with CANN 9.0.0	Complete tests/test_fill.py: 64 passed, 6 skipped

Full `pytest` output (optional)

collected 70 items
64 passed, 6 skipped, 15 warnings in 1.27s

Benchmark / Performance Impact

N/A. This PR adds functional provider coverage and does not claim a performance improvement.

Notes for Reviewers

• The scalar overload keeps the backing double as an operator member because aclCreateScalar retains its address.
• The tensor-value overload requires a one-element tensor on the same Ascend device as the output.
• Validation used the focused operator build and complete fill test file; the repository-wide test suite was not run.