design: Add 0004-multimodal-i2t proposal

[sangminwoo]

Description

Add design doc for multimodal image-to-text evaluation support in strands-evals SDK.

Introduces MultimodalOutputEvaluator extending OutputEvaluator to enable MLLM-as-a-Judge evaluation for multimodal tasks, starting with image/document-to-text. The evaluator constructs multimodal prompts using strands SDK ContentBlock format and supports both reference-free and reference-based evaluation with automatic rubric selection across four dimensions: Overall Quality (P0), Correctness (P0), Faithfulness (P1), and Instruction Following (P1).

Key design decisions:

• Extends OutputEvaluator with same Agent.__call__ invocation pattern (accepts both str and list[ContentBlock])
• Automatic reference-based rubric selection via _select_rubric() when expected_output is provided
• InputT=MultimodalInput (TypedDict) carries {"media": ImageData/AnyMediaData, "instruction": str} (modality-generic naming for future extensibility)
• ImageData supports file paths, base64, data URLs, HTTP URLs (auto-fetched via urllib.request), S3 URIs (auto-fetched via boto3), bytes, and PIL Images
• Built-in rubric templates + convenience subclasses per dimension

Related Issues

• strands-agents/evals Issue #128

Type of Change

• New content

Checklist

• I have read the CONTRIBUTING document
• My changes follow the project's documentation style
• I have tested the documentation locally using npm run dev
• Links in the documentation are valid and working

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[sangminwoo]

Hi @afarntrog, this PR is ready for the final review. I've updated the design doc to reflect your comments: 1/ added support for remote URIs/URLs and 2/ broadened the multimodality class to accommodate future media types. Would appreciate an approval when you get a chance so we can get this merged.