Update SVM + SVR datasets and unit tests

[jeff1evesque]

We will update our SVM and SVR related datasets, with more meaningful data. Specifically, we'll replace all corresponding dataset instances with the spambase dataset. We can use the full dataset for the data_new case, and a randomized subset, for the data_append dataset cases. Then, we'll need to adjust our unit tests, to be predicated on training with a randomized 2/3 of the dataset, then using the remaining 1/3 data to test, and compare that the results are within an acceptable tolerance.