Split multimodal group into vision and audio

[sjmonson]

Summary

Turns the guidellm[multimodal] extras group into guidellm[audio] and guidellm[vision].

---

• "I certify that all code in this PR is my own, except as noted below."

Use of AI

• Includes AI-assisted code completion
• Includes code generated by an AI application
• Includes AI-generated tests (NOTE: AI written tests should have a docstring that includes ## WRITTEN BY AI ##)

[Copilot]

Pull Request Overview

This PR splits the guidellm[multimodal] extras group into separate guidellm[audio] and guidellm[vision] extras to allow more granular installation of dependencies. Audio-related functionality is extracted into a new audio.py module, while vision-related functionality remains in vision.py.

Key Changes:

• Audio functionality moved from vision.py to new audio.py module
• Package extras split from single multimodal to separate audio and vision groups
• Import statements updated to reflect new module structure

Reviewed Changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated 3 comments.

File	Description
src/guidellm/extras/vision.py	Removed audio-related imports, functions (encode_audio, _decode_audio, _encode_audio), and updated error message to reference guidellm[vision]
src/guidellm/extras/audio.py	New module containing audio encoding/decoding functions moved from vision.py, including encode_audio, _decode_audio, _encode_audio, is_url, and get_file_name
src/guidellm/data/preprocessors/formatters.py	Updated import statements to reference new module locations (audio and vision)
pyproject.toml	Split multimodal extras into separate audio and vision groups with appropriate dependencies

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[jaredoconnell]

I haven't tested it, but the code change looks good. The only thing I want to make sure is tested is that it works correctly with only one of vision or audio and not both in terms of handling the import errors.