Image reader and image support in gather_evidence

[jamesbraza]

This PR is the first step in PaperQA becoming multimodal:

1. Adds reader support for images
2. Converts Docs to also store images (with text and metadata) as a ParsedMedia
3. Expands gather_evidence tool to include images in the Context-generation prompt
4. Adds tests of the prior three steps

[Copilot]

Pull Request Overview

This PR introduces multimodal capabilities to PaperQA by adding support for images as a new media type. The key changes enable the system to parse, store, and utilize images alongside text content in the question-answering process.

• Adds image parsing functionality with a new ParsedMedia class for storing image data and metadata
• Extends the Text class to include associated media and updates the evidence gathering process to incorporate images
• Integrates image support into the LLM prompting system for multimodal question answering

Reviewed Changes

Copilot reviewed 8 out of 12 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/paperqa/types.py	Introduces ParsedMedia class for image storage and extends Text with media field
src/paperqa/readers.py	Adds parse_image function and updates chunking logic to handle images
src/paperqa/core.py	Modifies evidence summarization to include images in LLM prompts
src/paperqa/prompts.py	Updates prompt templates to support multimodal content with image integration
src/paperqa/utils.py	Adds utility functions for base64 encoding/decoding of image data
src/paperqa/docs.py	Updates document addition logic to handle image parsing metadata
src/paperqa/settings.py	Excludes image files from indexing until embedding support is added
tests/test_paperqa.py	Comprehensive tests for image parsing, storage, and multimodal querying

Comments suppressed due to low confidence (1)

tests/test_paperqa.py:1338

• This test is comparing an object with itself, which will always be true. Consider testing equality with a separate instance that has the same content to properly test the __eq__ method.

    assert parsed_image == parsed_image, "Expected equality"  # noqa: PLR0124