[Feature requests] Precision refinements about rforestry

[theo-s]

I have implemented the scaling and centering of the data before we do C++ computations. This is in version .66. It included more changes than I initially thought, as many auxiliary functions also make use of the data as well, so we have to be careful to scale/rescale in these functions as well. I have tested this version, and for all of the current tests we have, it matches in performance with the unscaled version.

We expect that if we train a forest with scaling and without, the results will be the same for a fixed seed. The results are slightly different for the high precision tests, but are still deterministic. We expect this as once the features/outcomes are scaled, we do different floating point computations and this results in very small changes.