Richards benchmark test fails due to assert in spline

Probably something with the regularization of krw/pc-sw. The regularized interval approximated by the spline might be too small? It could also be that the assert is wrong. Needs closer investigation.