Add Qwen3.5 MoE (35B-A3B) model export and runner for CUDA backend

[mergennachin]

Memory-efficient loading using meta-device construction + lazy
safetensors shard-by-shard loading + assign=True state dict loading,
following the voxtral_realtime pattern. Peak CPU memory during loading
is ~1x model size instead of ~3x.

Expert weights are structured as grouped nn.Linear modules (16 groups
of 16 experts each) so quantize_model_() handles them automatically.
Layer-by-layer quantization on CUDA avoids loading the full bf16 model
onto GPU at once.

Includes C++ runner using the shared TextLLMRunner, Makefile target,
and CMake presets.

Reference implementations:

• https://github.com/mergennachin/nano_qwen35_moe/
• vLLM: vllm/model_executor/models/qwen3_5.py

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/18169

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 1 Unrelated Failure

As of commit 78a940d with merge base e458023 ():

NEW FAILURES - The following jobs have failed:

• Build Cadence / cpu-x86 / build (gh)
  RuntimeError: Command docker exec -t 08e8ae112d897801d08dc544bfa97f83d3085e6679451f9c8d746eefceb15d81 /exec failed with exit code 1
• Lint / link-check / lint-urls (gh)
  Process completed with exit code 1.

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• Test CUDA Windows Export and E2E / export-model-cuda-windows-artifact (mistralai, Voxtral-Mini-3B-2507, non-quantized) / linux-job (gh) (detected as infra flaky with no log or failing log classifier)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.