Add tests for multi-image VLM prompts, and for followup prompts

[neilmehta24]

We have tests for single image prompts and text-only prompts in test_vision_models.py. We should add tests for continued conversations which may have multiple images as part of the conversation.

[YorkieDev]

Should be solved by #67