Add Qwen 3.6 MoE model and switch CI to Qwen3.6-35B-A3B-HQQ-INT4 (#18… #49

Job	Run time
export-model-cuda-windows-artifact (nvidia, parakeet-tdt, non-quantized) / linux-job	1d 0h 0m 0s
export-model-cuda-windows-artifact (mistralai, Voxtral-Mini-3B-2507, quantized-int4-weight-only) / linux-job	1d 0h 0m 0s
export-model-cuda-windows-artifact (mistralai, Voxtral-Mini-3B-2507, non-quantized) / linux-job	1d 0h 0m 0s
export-model-cuda-windows-artifact (facebook, dinov2-small-imagenet1k-1-layer, non-quantized) / linux-job	1d 0h 0m 0s
export-model-cuda-windows-artifact (mistralai, Voxtral-Mini-4B-Realtime-2602, quantized-int4-tile... / linux-job	1d 0h 0m 0s
export-model-cuda-windows-artifact (nvidia, parakeet-tdt, quantized-int4-weight-only) / linux-job	1d 0h 0m 0s
test-model-cuda-windows-e2e	0s
	6d 0h 0m 0s

Provide feedback