openrct2_theme_park_ai score

Hello! Thanks for the great work!

I have a quick question regarding the `openrct2_theme_park_ai` task. The judge clamps each scenario’s value ratio at 2.0 and reports `avg(min(ratio, 2.0)) × 10000`, so the highest raw score it can emit is 20,000. However, the task’s rescale spec maps 0–40,000 → 0–100, so `score_0_100` can never exceed 50, and any agent that saturates the judge is reported as 50/100. Should the reported score therefore be multiplied by 2?

Start by locating the openrct2_theme_park_ai judge and its rescale specification. Check how the clamped ratio and 0–40,000 mapping produce score_0_100, then verify the reported score reaches the intended 0–100 range without changing unrelated task scoring.

Written by the indexing model from the issue text.