Add Gemma2-27b-q8 benchmark result to leaderboard

[youknow04]

Adding Gemma2-27b-q8 result to leaderboard for local LLM enthusiasts.

• Tested with Ollama 0.2.1
• gemma2:27b-instruct-q8_0 model with hash dab5dca674db

This LLM model struggles with diff formatting; whole format works better.

The quality is not as good as rumored, at least for coding with Aider. Those with high expectations might be disappointed.

[paul-gauthier]

Thanks for running and submitting this!

[paul-gauthier]

Thanks!