Add support for VibeVoice

[IgorSwat]

I was trying to export the VibeVoice Text to Speech model with the optimum-cli export script, but it fails to recognize the model and results in NoneType object error, no matter which task option do I choose. This is the command I have been using:

optimum-cli export executorch --model microsoft/VibeVoice-Realtime-0.5B --output_dir . --recipe xnnpack --task text-to-speech --dtype float32

I would be grateful for support for VibeVoice in optimum-cli, or pointing out the correct task option if it is already supported, so the model can be exported without manual experimentation.