Rt models (with two lithologies) must have contact layer specified as an input file

[kbarnhart]

Historically all terrainbento models were initialized using a DEM, thus we specified the contact layer as a DEM.

However, for synthetic test cases, it would be nice to be able to specify the contact elevation as a numpy array.

[kbarnhart]

An additional thought here.

This issue can be generalized as a problem with the rock till contact, the initial topographic elevation, AND the initial soil thickness.

A possible approach would be to permit that each of these three fields can be specified either by an input file OR by parameters to a generic function "create_field"

it could take as arguments options related to the mean of the initial surface (as a function of node_x and node_y, where this mean is applied (core nodes or all nodes), the properties of random noise applied, and where this noise is applied.

This may also help with the standardization of a BMI for terrainbento