[Feature Request]: Refactor CRITIC

[alckasoc]

Feature Description

The current CRITIC implementation has an if-statement condition for every type of benchmark. As of writing this, it has a check for "search" (QA benchmarks) and "code_interpreter" (math reasoning benchmarks). How can we refactor the generate() method to generalize across these benchmarks so we don't have a condition for each category of benchmark?

Reason

This will help streamline the addition of future benchmarks.