feat: implement Google Batch Mode support for LLM calls

[devin-ai-integration]

Fix Google Batch Mode Type Errors and API Compatibility

Summary

This PR fixes 3 critical type-checker errors in the Google Batch Mode implementation for CrewAI's LLM class. The errors were caused by using genai.Client() which doesn't exist in the current google-generativeai SDK v0.8.5.

Key Changes:

• ✅ Fixed type-checker errors by replacing genai.Client() with genai.GenerativeModel()
• ✅ Implemented sequential processing fallback since current SDK lacks batch API
• ✅ Added proper type annotations for batch result storage
• ✅ Maintained same user interface and event-driven architecture
• ✅ All custom batch mode tests pass

⚠️ Important: This implementation uses sequential processing instead of true batch API calls, as the current Google GenAI SDK doesn't support batch operations. This maintains the user interface but may not achieve the same cost optimization as true batch processing.

Review & Testing Checklist for Human

🔴 High Priority (3 items)

• Test with real Google API credentials - Verify genai.GenerativeModel() calls work correctly with actual API keys (my testing was mocked)
• Validate cost optimization expectations - Sequential processing may not achieve the 50% cost reduction mentioned in issue [FEATURE] Support Google Batch Mode for LLM calls #3116. Confirm if this fallback approach is acceptable
• Verify batch mode user experience - Test the complete flow from user perspective to ensure the interface works as expected despite internal changes

---

Diagram

%%{ init : { "theme" : "default" }}%%
graph TB
    Issue["`GitHub Issue #3116<br/>Google Batch Mode Support`"]
    
    subgraph Core["Core Implementation"]
        LLM["`src/crewai/llm.py<br/>LLM Class`"]:::major-edit
        BatchMethods["`Batch Methods:<br/>_submit_batch_job()<br/>_poll_batch_job()<br/>_retrieve_batch_results()`"]:::major-edit
    end
    
    subgraph SDK["Google GenAI SDK"]
        OldAPI["`genai.Client()<br/>(Non-existent)`"]:::context
        NewAPI["`genai.GenerativeModel()<br/>(Current SDK)`"]:::major-edit
    end
    
    subgraph Testing["Testing"]
        TestRunner["`test_runner.py<br/>Custom Test Suite`"]:::minor-edit
        BatchTests["`tests/test_batch_mode.py<br/>Comprehensive Tests`"]:::context
    end
    
    Issue --> LLM
    LLM --> BatchMethods
    OldAPI -.-> |"Fixed Type Errors"| NewAPI
    BatchMethods --> NewAPI
    TestRunner --> BatchMethods
    BatchTests --> BatchMethods
    
    subgraph Legend
        L1[Major Edit]:::major-edit
        L2[Minor Edit]:::minor-edit  
        L3[Context/No Edit]:::context
    end
    
    classDef major-edit fill:#90EE90
    classDef minor-edit fill:#87CEEB
    classDef context fill:#FFFFFF

Loading

Notes

• The implementation uses sequential processing as a fallback since the current Google GenAI SDK v0.8.5 doesn't have a Client class or batch API support
• All type-checker errors are fixed and CI passes, but real-world testing with actual API keys is recommended
• The implementation maintains the same user interface and event emission, making it compatible with existing code
• This approach provides a foundation for future integration with true batch API when it becomes available in the Google GenAI SDK

Link to Devin run: https://app.devin.ai/sessions/a16f00eb331b443fb7f62d4448a8453f
Requested by: João (joao@crewai.com)

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring