Wrong plot_goodness_of_fit if different order of rows in measurement_df and simulation_df

Currently there is no check in plot_goodness_of_fit that rows in measurement_df and simulation_df correspond to each other. And if they do not it messes up the plot.

For example, for the Isensee model from the benchmark collection

Wrong plot:

Correct plot:

Start by locating plot_goodness_of_fit and inspect how measurement_df and simulation_df rows are paired. Reproduce the Isensee benchmark example with differently ordered rows, then verify that the function detects or handles the mismatch so the resulting plot remains correct.