[FEATURE] Support local question answering model

[TrungBui59]

Is your feature request related to a problem? There is a demand to support the question-answering model to ml-commons. Since there is already a PR for the question-answering model in the py-ml repo. By supporting this type of model, we can roll out the feature right away to the user

Input: list of string Output: List of string

What solution would you like? Following the same approach as the text embedding model, we will let the user upload and run the model on top of DJL

What alternatives have you considered? I haven't figured out an alternative approach. But this approach reuses the approach we did for text embedding, which is proven to be working

Do you have any additional context? HuggingFace tutorial