Support multi-modal input from file for OpenAI Chat Completions

[nv-hwoo]

The PR allows users to provide a file that contains multi-modal data (in this case, texts and images) which will be used to generate input payloads for OpenAI Chat Completions.

For instance, when the user provides a file with multi-modal data:

{ "text_input": "prompt1", "image": "<path>/<to>/<image1>"},
{ "text_input": "prompt2", "image": "<path>/<to>/<image2>"},
...

GenAI-Perf will create a multi-modal input payload as following:

{
    "message": [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "prompt1",
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "data:image/png;base64,...",
                    }
                },
            ]
        }
    ],
    "model": "some-model"
},
{
    "message": [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "prompt2",
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "data:image/png;base64,...",
                    }
                },
            ]
        }
    ],
    "model": "some-model"
},
...

[dyastremsky]

Great work!