feat(converter): add video converter.

[absadiki]

This PR is a step towards resolving #154.

It introduces a VideoConverter class that converts videos to markdown by:

• Extracting metadata (if exiftool is installed)
• Performing speech transcription (if speech_recognition and pydub are installed)
• Generating a summary via a multimodal LLM from the transcription [This is optional and defaults to True if llm_client is configured]

Notes:

• I believe checking the file type based on the extension is not ideal. There are many video extensions, and I think checking the mime_type would be a better approach, as it can cover a wider range of video files.
• I’m unsure about the testing strategy .. should we focus only on testing exiftool? Please share your thoughts on this.
• Additionally, I suggest refactoring Mp3Converter into a more general AudioConverter, as there are many audio extensions to consider. If you agree with this, I can submit a separate PR for it.

[l-lumin]

could you add tests?

[absadiki]

@l-lumin, could you provide a sample video file that is allowed to be uploaded to the repo?