Add primitives to build and manipulate EntitySets inside a pipeline

[csala]

• MLPrimitives version: 0.1.6

Description

Current featuretools.dfs primitive is ready to work using an EntitySet. However, at the moment the only possibility is to pass the EntitySet as an argument to the pipeline fit and predict methods, so it is not possible to use en EntitySet that contains the output of a previous primitive.

To solve this, two primitives will be added:

• featuretools.EntitytSet.entity_from_dataframe: Gets an entityset and adds a dataframe to it calling its entity_from_dataframe method, and finally returns the updated entityset. It accepts and passes along all the arguments that the method accepts. If the entityset is None, it creates a new one.
• featuretools.EntitySet.add_relationship: It gets and entityset and calls its add_relationship method passing the 4 required references.