Evaluator output should use a templating engine like Chunk

[logological]

Originally reported on Google Code with ID 28

We have various classes for evaluation and reporting which we subclass according to
the output format (text, HTML, etc.).  It would be simpler and more flexible if instead
we rewrote these classes to use a templating engine like Chunk: https://code.google.com/p/chunk-templates/

Reported by [email protected] on 2013-08-02 14:26:03