Add MingAudio class

[Deep-unlearning]

Add MingAudio wrapper class

This PR adds a MingAudio class to simplify model usage.
Previously, users had to define this class manually; it’s now directly available for import.

Highlights:

• Introduces MingAudio in modeling_ming_audio.py with MingAudioConfig.

• Enables one-line usage:

  from ming_audio import MingAudio
  model = MingAudio.from_pretrained("inclusionAI/Ming-UniAudio")

• Aligns the repo with the standard Hugging Face model interface for easier loading and extension.

[kingmpw2015]

Awesome contribution! Thanks for simplifying the usage of Ming-UniAudio. We'll review the code.