Support qwen3 moe

[lu4p]

When trying to load this model: https://huggingface.co/mlx-community/Qwen3-30B-A3B-4bit-DWQ

I get

🥲 Failed to load the model

Failed to load model

Error when loading model: ValueError: Model type qwen3_moe not supported.

[chef-andre]

Try this version from lmstudio-community:

https://huggingface.co/lmstudio-community/Qwen3-30B-A3B-MLX-4bit

I was having a similar issue with mlx-community/Qwen3-30B-A3B-8bit, but not this one: lmstudio-community/Qwen3-30B-A3B-MLX-8bit

My best guess about root cause is here https://huggingface.co/Qwen/Qwen3-30B-A3B#quickstart (transformers<4.51.0)