Refine Whisper and WeNet Contents Extractor

[Adorable-Qin]

• Deleted modules/whisper_extractor. Change to OpenAI's Whisper API for content features' extraction.
• Deleted redundant files in modules/wenet_extractor.

[RMSnow]

Are there any more pip dependencies?

[RMSnow]

We need to discuss for whisper extractor. Line154 is modified by me, which is different from the original whisper code. The purpose is to avoid from downloading whisper's pretrained model automatically from the OpenAI's server, since some running servers of users (like me) are out of network.

[Adorable-Qin]

If you put your checkpoints in the right place, i.e. model path specified in your config, I don't think it should be a problem though.

[Adorable-Qin]

Are there any more pip dependencies?

No. Since openai-whisper has already been added to env.sh, we don't need to borrow it to the repo as code format, nor do we need to add the other pip dependency.

[Adorable-Qin]

Modified the download_root of whisper model. If the given path is a file, its dir will be used as the download root. On the other hand, it will be used directly if it is a folder.

[RMSnow]

Use black to format this file.