Add ability to add arbitrary code to run evaluations

[AdamHess]

Describe the feature or potential improvement

instead of just offering LLM as judge add the ability to add arbitrary programming code to run the evaluation.

(ideally python, but TS makes more sense given the library is TS specific)

Additional information

No response

[jannikmaierhoefer]

Thanks for adding this @AdamHess. We have this on our roadmap here: https://github.com/orgs/langfuse/discussions/6087