[Feature] Image fields for multi-modal models

I'm trying to evaluate Claude v3's performance for some document understanding tasks, with a workflow that includes passing the image of the page in as one of the inputs.

Is fmeval considering native handling for image/multi-modal fields in input datasets?

The issue does not name any files, tests, or entry points. Start by locating how input datasets and model inputs are represented, then trace the Claude v3 evaluation path to determine where image or multi-modal fields would fit. Done would mean a defined, tested approach for passing image inputs through an evaluation workflow.

Written by the indexing model from the issue text.