add small set of test datasets to be used for regression testing

[john-waczak]

A simple regression dataset like dr. lary's nonlinear function from class as well as the iris dataset.

All of our big "curated" datasets can be added to OSN in a new folder, say "/ml-datasets"

[john-waczak]

e.g. Dr. Lary's test function:

$f(x,y) = x^5 + y^3 - x^4 - y^3$

[john-waczak]

iris dataset, mnist, etc...

[john-waczak]

colors dataset for unsupervised classification methods e.g. KNN, DBSCAN, SelfOrganizingMaps, GTM, etc...

[john-waczak]

We should add some test datasets with missing values, nans, infs, etc to test the pipeline