image content support

[mr-brobot]

i realize this is complicated by the fact that llm providers do not agree on a standard image message format

perhaps the mcp client can accept a conversion function image_content_to_message(content: ImageContent) -> HumanMessage | AIMessage that allows the user to control this

i'm happy to implement, but curious if anyone has a better approach in mind

[vbarda]

@mr-brobot might be a bit hard to standardize, but i think we do need to support properly. i have a TODO to add a PR for this

[vbarda]

We're actually adding support for a generic format for multi-modal message content in langchain libraries, so I will wait until that is released and then add in this lib - stay tuned!