Nice! Let's switch to PCHIPInterpolation for this and #2446. I think you also need to add Plots to the root environment, though I hope that won't undo the gains from #2441 too much.

Turns out PCHIPInterpolation cannot deal with only 2 data points:

SciML/DataInterpolations.jl#446

Will do a quick fix on our side with a reference to the issue

Of course some simulation results changed due to this interpolation type change, I guess the benchmark simulation results for the integration and regression tests have to be updated as well