Store labeled pairs in a table

[ecsalomon]

Labeled training example pairs should be stored in a table for selection and reuse. Data stored for examples should include:

• Source
• Source ids
• Label
• Label date
• Comment to store labeling rules applied by labeler

Storing examples like this allows them to by reused in the following ways:

• Select specific subsets of labeled pairs to build models from
• Store multiple labels for the same pair to determine how different labeling rules affect results
• Dynamically format labeled examples for models that use different set of input data by linking examples to source data tables

This will also entail a test that any model is only trained on one label per pair!

The set of ids of these pairs should be used in generating model ids (#18).

This issue is the same as dssg/pgdedupe#63.

[jtwalsh0]

Use a timestamp rather than date

[ecsalomon]

We are using unsupervised learning