Add retail support evaluation cookbook using Langfuse Datasets and experiments

[hienhtphan]

Hi Langfuse team,

I am posting from hienhtphan, admin of the SData-Pro GitHub organisation, and would submit the PR from the SData-Pro/langfuse-docs fork.

I would like to contribute a small cookbook notebook to langfuse/langfuse-docs under cookbook/, following the docs contribution flow and notebook conversion step described in the repo README.

The proposed notebook would use a tiny inline synthetic retail support dataset to demonstrate Langfuse Datasets, JSON Schema validation, run_experiment, deterministic evaluators, trace/dataset-run scores, and a simple CI-compatible regression gate. It would avoid external data files, warehouse-specific setup, paid/private data, and vendor-specific product placement; the goal is a copy-pasteable evaluation workflow for support-style LLM applications.

Would this be a useful contribution, and is cookbook/evaluation_retail_support_dataset.ipynb the right location and filename before I draft the PR?

Should contributors run bash scripts/update_cookbook_docs.sh and include the generated output in the PR, or do maintainers handle that step?