Parse and score the Judge's output

The Judge outputs JSON with scores for 5 criteria (Clarity, Feasibility, Impact, Innovation, Adherence), but it's still a raw string. You need to:

1. Parse the JSON — Convert the string into a usable object
2. Clean up formatting — Remove any markdown code blocks the AI might have added
3. Calculate the average score — Determine overall quality
4. Add final_score — This is what the IF node will check

This Code node is the bridge between the Judge's output and the decision logic.