Save dataset with model

[djross22]

When a model gets initiated, it would be good if the dataset information is saved with it in an easily accessible way, so that the model can be shared between different people, using, for example, a pickled model file.

Then, the user who receives the pickle file, can reconstitute the original model object easily and with minimal risk of error.

In current practice, to share a model, the user who initiates the model passes the code used to generate the model (or assumes that the receiver already has a copy of the same code). This can lead to subtle errors. For example, if the order of the phenotypes (a list of strings) is different between the code use by the initiator and the receiver, the phenotype information will get scrambled, and the receiver may end up inadvertently/unknowingly comparing predictions for one phenotype with experimental values for a different phenotype.

So, the easiest way I could find to do this is to save the dataset object as an attribute of the surface attributer of the model. And, since the surface attribute is typically an object of type lantern.model.surface.Phenotype, the only place I could find to easily make this change was in the Phenotype class constructor methods.

[djross22]

I updated the tests, so now the revised code passes. I'll approve after a couple of days if no one raises any objections.