Optimize shared code between explainers

[zyteka]

The explainers of a single RealApp object all need to transform the dataset separately, even though this is often the same process between explainers. We should investigate ways to share this process, while recognizing that different explainers may require different algorithm feature spaces. One option is to switch explainers to explicitly use the model-ready space whenever possible, and then have RealApp objects store the model-ready dataset for each model.

Another option for handling this involves caching the results of the transform_to_x_... functions