Meta-learner prediction method and testing set up

[kyle-messier]

No description provided.

[kyle-messier]

combining issue #38 #39 #40 #41 #42 #43 #44 #45 #47 #48

pull request #166 is open. The goal is to write a clean function for predicting the BART meta-learner to locations in SpatRast (netCDF) and sf-point formats. I have been trying to get the "predict" methods from BART, terra, sf to work, but am getting unknown method errors. Hopefully someone can have better luck.

Keep in mind this only has to be a minimally working version - it will be refactored later.

[kyle-messier]

@sigmafelix I'll pass this over to you. Please have @eva0marques help with prediction and/or corresponding testing function.

[kyle-messier]

@eva0marques @sigmafelix To clarify, the goal here in the short-term is to get a minimally working meta_learner_predict function that takes the BART model and prediction location covariates - outputs the prediction results as SpatRast and sf options. The remaining tests will hopefully be easy to do once we get this function written.