Add a speech-to-video workflow for MuseTalk (hosted on HF)

[jbilcke-hf]

MuseTalk is a model that was published some months ago (see https://huggingface.co/spaces/TMElyralab/MuseTalk) but it is not easy to use from an API.

I need to work on the automatic hugging face spawner (to automatically deploy and bill spaces) but until I have time to do this, I think we can just reserve the workflow for private use, by doing this:

1. add a workflow for MuseTalk (I will use my private fork https://huggingface.co/spaces/jbilcke-hf/MuseTalk)
2. add an environment variables for allow/reject lists of workflow IDs
3. make sure we properly hide the MuseTalk workflow if it is in the rejection list

[jbilcke-hf]

Well actually there is also a workflow on Replicate, it doesn't seem very fast but I've added it anyway

For the demo I still need the HF-hosted MuseTalk though, on a large GPU