Use more complex data in test_roc_curve_display.py #31243
Description
data_binary fixture in test_roc_curve_display.py, which uses the iris dataset:
scikit-learn/sklearn/metrics/_plot/tests/test_roc_curve_display.py
Lines 28 to 30 in af7df5c
always gives a AUC value of 1.0 (with and without drop_intermediate). Should we consider using more complex data?
Noticed in #30399 (comment)
cc @glemaitre