Refactor Message Struct: Remove MultiContent, Add UserInput and AssistantOutput for Multi-modal Support

[zhuangjie1125]

📌 Summary

To support multi-modal input and output for the model, Eino plans to make the following changes to the Message struct:

Changes

• ✅ New Features

  • Added Fields:
    • UserInputMultiContent []ChatMessageInputPart : This field captures the multi-modal input provided by the user.
    • AssistantGenMultiContent []ChatMessageOutputPart : This field captures the multi-modal output generated by the assistant.

• ⛔ Deprecated (will be removed in future releases)

  • Removed Fields:
    • MultiContent []ChatMessagePart : This field has been removed as it is no longer maintained.
    • ChatMessageImageURL, ChatMessageAudioURL, ChatMessageVideoURL, ChatMessageFileURL: These types have also been removed to simplify the structure and focus on multi-modal handling.