What does this PR do?

This PR adds a dedicated Whisper Model Leaderboard tab to compare different Whisper model implementations across various backends and hardware configurations.

Key Changes:

• Added new environment variables for Whisper-specific repository and paths
• Extended load_all_info_from_dataset_hub() to handle both original and Whisper evaluation results
• Created a new "🔄 Whisper Model Leaderboard" tab with:
  • Additional columns for Backend and Device information
  • Interactive filters to compare results by backend type and hardware
  • Same evaluation metrics (WER, RTFx) as the main leaderboard

Why?

This dedicated view will help users:

• Compare performance across different Whisper implementations
• Understand speed/accuracy tradeoffs on different hardware
• Make informed decisions about which backend to use for their specific needs

Testing

• Verified that both original and Whisper leaderboards load correctly
• Tested filtering functionality with multiple backend/device combinations
• Confirmed that all WER and RTFx metrics display properly

Screenshots