Fix CUDA configuration issue #1140 - add better error handling and fa…

[maheshreddyy345]

This PR addresses issue #1140 where users with misconfigured CUDA environments encounter the error "CUDA_PATH is set but CUDA wasn't able to be loaded".

Changes Made

• Added comprehensive error handling for CUDA initialization failures
• Implemented automatic fallback to CPU when CUDA is unavailable but requested
• Added detailed diagnostics for CUDA configuration issues
• Improved error messages to help users troubleshoot CUDA setup problems

Technical Details

• Enhanced decode_execution_providers() in core.py to detect and validate CUDA configurations
• Added fallback logic in face_swapper.py and face_analyser.py to gracefully handle CUDA initialization failures
• Created a CUDA diagnostics function that provides useful troubleshooting information
• Maintained full backward compatibility with existing behavior when CUDA is properly configured

Testing

These changes have been tested in environments with various CUDA configurations:

• Working CUDA setup (no changes to behavior)
• Missing CUDA installation (graceful fallback to CPU)
• Misconfigured CUDA environment variables (clear error messages and CPU fallback)

The application now runs at least 15 minutes continuously with these changes, meeting the project's stability test requirements.

User Experience Improvements

Users who encounter CUDA issues will now:

1. See helpful diagnostics about their CUDA configuration
2. Automatically continue running with CPU rather than experiencing crashes
3. Receive clear guidance on how to fix their CUDA setup

This fix improves the overall robustness of the application, particularly for users who are less familiar with configuring CUDA environments.

Branching Strategy

This PR is submitted to the premain branch as specified in the contribution guidelines.

Fixes #1140

Summary by Sourcery

Improve CUDA configuration handling and error resilience by adding comprehensive error detection, diagnostics, and automatic fallback mechanisms for CUDA initialization failures.

New Features:

• Added CUDA configuration diagnostic checks to help users troubleshoot environment issues

Bug Fixes:

• Fixed issues with CUDA initialization by adding robust error detection
• Prevent application crashes when CUDA is misconfigured

Enhancements:

• Enhanced error handling for CUDA provider initialization
• Implemented automatic fallback to CPU when CUDA is unavailable
• Added detailed logging for CUDA configuration problems