Pareto plotting fails with KeyError: "['ds'] not found in axis" during evaluate_models

Summary:
After Pareto fronts are computed, the evaluation crashes in the plotting data generation step with KeyError: "['ds'] not found in axis". Optimization finishes, but building plot data fails.

What I expected:
Evaluation completes and Pareto plots are produced.

What happened:
Pareto optimization completes, then plot_data_generator errors out with KeyError: "['ds'] not found in axis" (logs below).

Notes / hypothesis:
This seems to happen if the date field ds isn’t present as a column (e.g., it’s on the index or named differently). If required, could the library either (a) auto-handle ds on the index or (b) raise a clearer early validation error?

Question:
Is ds strictly required to be a column (not index) for Pareto plotting? If so, can this be validated earlier or handled automatically?