[Feature] adding code for groq support

[rushabh31]

What does this PR do?

This PR adds support for Groq Vision API and Vertex AI (Google Cloud) in the vision-parse library, enabling users to leverage multiple powerful vision models for processing images and PDFs.

• Adding support for Groq API for vision models
• Adding support for Vertex AI integration for vision models

Features Added

1. Groq Integration: Implemented full support for Groq's vision models (meta-llama/llama-4-scout-17b-16e-instruct and meta-llama/llama-4-maverick-17b-128e-instruct)

2. Vertex AI Integration: Added support for Google Cloud's Vertex AI platform with Gemini models (gemini-1.5-pro-002 and gemini-1.5-flash-002)

3. Flexible Configuration Options:

   • Added groq_config parameter to both LLM and VisionParser classes
   • Added vertex_config parameter with support for multiple authentication methods (API key, service account JSON, service account key file)

4. Robust Error Handling:

   • Implemented specific error handling for Groq's pixel size limitations
   • Added error handling for Vertex AI image size and dimension constraints
   • Providing clear guidance to users when images exceed API limits

5. Documentation & Examples:

   • Created example scripts demonstrating both Groq and Vertex AI API usage
   • Added test examples for verifying integrations
   • Updated dependency management (pyproject.toml) with appropriate requirements

6. Performance Optimization: Added guidance for proper image resolution settings to stay within API limitations

7. Page-Level Visual Analysis: Implemented a new workflow to send entire page images to LLMs for detecting and summarizing embedded visuals like images, diagrams, charts, and visualizations

8. Configurable Visual Summary: Added enable_image_summary parameter to toggle visual element detection and summary generation

Implementation Details

Groq Integration:

• Added Groq models to supported models in constants.py
• Extended LLM class to include Groq client initialization and request handling
• Updated VisionParser to accept and pass through Groq configuration
• Added proper error detection and messaging for Groq-specific limitations
• Added proper optional dependency for Groq in pyproject.toml

Vertex AI Integration:

• Added Vertex AI models to supported models in constants.py
• Implemented Vertex AI client initialization with multiple authentication methods
• Added _vertex method to handle image processing through Vertex AI
• Updated VisionParser to accept and pass vertex_config parameter
• Added comprehensive error handling for Vertex AI limitations
• Created usage examples for Vertex AI integration
• Added proper optional dependencies for Vertex AI in pyproject.toml

Page-Level Visual Analysis:

• Implemented new workflow to send entire page images to LLMs for visual element detection
• Added page_visuals_prompt template for instructing LLMs to identify and summarize embedded images, charts, diagrams, etc.
• Created detect_page_visuals method in LLM class to handle the visual detection and summarization process
• Updated parser workflow to integrate visual summaries into the markdown output
• Removed legacy individual image extraction code in favor of the more efficient page-level approach

Configurable Visual Summary:

• Added enable_image_summary parameter to LLM class with default value of True
• Extended VisionParser to accept and pass this parameter to LLM instance
• Updated the conversion logic to conditionally perform visual analysis based on the parameter value
• Added example usage in documentation to demonstrate how to toggle the feature

Before submitting

• This PR improves the library by adding support for new LLM providers (Groq and Vertex AI)
• Ran make lint and make format to handle lint / formatting issues
• Ran make test to run relevant tests scripts
• Read the contributor guidelines
• Wrote example code demonstrating the new functionalities
• Added tests for both Groq and Vertex AI integrations

Testing

Groq Testing

• Manually tested with the Groq API to verify the implementation works correctly
• Created unit tests for both the LLM and parser classes to ensure proper integration with Groq
• Verified error handling for image size limitations

Vertex AI Testing

• Added tests for Vertex AI client initialization with different authentication methods
• Created unit tests for both the LLM and parser classes to ensure proper integration with Vertex AI
• Verified proper error handling for Vertex AI-specific limitations
• Created example script demonstrating usage of Vertex AI with proper configuration

[rushabh31]

@iamarunbrahma Added support for Groq vision models.