Add Multi-GPU Support for LlamaCpp Engine

[nguyenhoangthuan99]

Add Multi-GPU Support for LlamaCpp Engine

Description

We need to implement multi-GPU support for our LlamaCpp wrapper engine to improve performance and allow users to utilize multiple GPUs effectively.

Goals

• Allow users to choose which available GPUs to use for running the engine
• Implement load balancing across selected GPUs
• Maintain compatibility with single-GPU setups

Proposed Implementation

1. Detect available GPUs on the system
2. Add a configuration option for users to specify which GPUs to use
3. Modify the wrapper engine to distribute workload across selected GPUs

Acceptance Criteria

• [ ] Users can specify which GPUs to use via configuration
• [ ] The engine correctly utilizes all selected GPUs

Additional Considerations

• Ensure proper error handling for scenarios where specified GPUs are unavailable
• Will we add this feature to model.yml for model management?
• Is this feature works for both CLI and API?