feat: add multimodal support to AnthropicChatGenerator

[ChinmayBansal]

Related Issues

• fixes Image support in AnthropicChatGenerator #2125

Proposed Changes:

This PR adds multimodal support to AnthropicChatGenerator, enabling it to handle both text and image inputs
in user messages. The implementation follows the same patterns established by the HuggingFace and Amazon
Bedrock integrations.

Key changes:

• Modified _convert_messages_to_anthropic_format() to handle ImageContent alongside TextContent
• Added proper Anthropic vision API format using ImageBlockParam with base64 image encoding
• Implemented content order preservation for mixed text/image messages
• Added validation to prevent images in assistant messages (Anthropic API limitation)
• Updated component docstring with multimodal usage example
• Added comprehensive type safety with proper Anthropic types

Technical details:

• Uses ImageBlockParam with proper media type casting for Anthropic's vision API
• Supports all Anthropic-compatible image formats: JPEG, PNG, GIF, WebP
• Maintains backward compatibility with existing text-only functionality
• Integrates seamlessly with existing prompt caching and tool calling features

How did you test it?

Unit Tests:

• ✅ All existing unit tests pass (49 passed, 4 skipped)
• ✅ Added test_convert_message_to_anthropic_format_with_image to verify proper message conversion
• ✅ Added validation test to ensure images in assistant messages raise appropriate errors

Integration Tests:

• ✅ Added test_live_run_multimodal for real API testing (requires ANTHROPIC_API_KEY)
• ✅ Uses test image from shared test assets (apple.jpg)

Code Quality:

• ✅ Linting passes (hatch run fmt)
• ✅ Type checking passes (hatch run test:types)
• ✅ Follows existing code patterns and conventions

Manual Verification:

• Tested with Claude Sonnet 3.5 using various image types
• Verified proper error handling for unsupported scenarios
• Confirmed multimodal messages work with different content ordering

Notes for the reviewer

• The implementation closely follows the Amazon Bedrock integration and HuggingFace integration patterns for consistency
• Image validation ensures only user messages can contain images (Anthropic API requirement)
• Type annotations use proper Anthropic ImageBlockParam types for full type safety
• Test image (apple.jpg) is shared from Bedrock integration's test assets
• The cast() function is used for media type conversion to satisfy Anthropic's strict typing

Checklist

• I have read the contributors
  guidelines and the
  code of conduct
• I have updated the related issue with new insights and changes
• I added unit tests and updated the docstrings
• I've used one of the conventional commit types for
  my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test:.

Suggested PR Title (using conventional commits):
feat: add multimodal support to AnthropicChatGenerator

[CLAassistant]

All committers have signed the CLA.

[anakin87]

Thanks for this contribution!

This PR is already good. I left some comments.

[anakin87]

Thanks!