Support base64 images in Function Tool Call Responses

[BenConstable9]

What feature would you like to be added?

Support for multi-modal messages in a Function Tool Call response similarly to the MultiModal input message.

Why is this needed?

It allows agents to call tools that will return images e.g. part of a multi-modal rag scenario

[victordibia]

Hey @BenConstable9 ,

Is this possibly related to #4725 ?
Any chance you are interested in contributing to this? We can help along the way.

In the mean time, it will be a huge help if you can describe the scenario you have in mind in detail. E.g, are you using core api vs agentchat? Can you give a general example of the the code/api you expect, how you would consume the multimodal message (or other message types etc).