[Feature] Adding manual derivate creation as predictor helper function

[Martin-Jung]

Although it is already possible to manually create and alter derivates (predictor_transform() or add_predictors()), it might in many cases be beneficial to manually control this process much more.
For example:

• When a specific threshold (50% forest cover) needs to be used.
• A specific function (complex spline or similar) is to be applied to a single or multiple predictors.

Thus I propose to add a add_predictors_prior() function that is customized to adding controlled derivates to a distribution model. Alternatively specific thresholds could be added as parameter to the extisting add_predictors() function so as to allow the specification of variable-specific derivates.

• Make sure the threshold and other derivate functions easily allow this
• Support custom formula based transformation
• Add add_predictor_derivate() function or similar to the package
• Make a unit test to see that everything works.

Thoughts welcome...