[V3] Fix the bug of text-generation example

[NingW101]

• Update the package name to huggingface/transformers
• Update model file name of distilgpt2
• Add TextStreamer to handle callback function for TextGenertaionPipeline

[NingW101]

I tried to test text-generation example with transformers-v3 and found that models could not be successfully fetched from HF because the path of onnx model was not correctly concatenated, besides, the code for dispatching callback function when new token is generated has been changed to use Streamer (including accept cb, decode the new generated tokens), however it has not been implemented in pipeline. I have tested and the example went well.

@xenova PTAL