Update Evaluation page

https://deeplearning4j.org/evaluation

(I'll probably need to have some input on this, just flagging it so it doesn't get forgotten)

> Precision, recall & F1: macro-averaged (equally weighted avg. of 10 classes)
> Precision, recall & F1: reported for positive class (class 1 - "1") only

We should probably explain what this means, and how to change the positive class for the binary case.
Also post next release, we've changed the confusion matrix format; it isn't printed by default for a very large number of classes. We should clarify how to override the default (and print it anyway).

A general review/polishing pass wouldn't hurt either.

## Due Date
*To be completed by:* YYYY-MM-DD

## Description
*Write a short description of what needs to be done.*

## Assignees
*Please ensure you have assigned at least one person to this issue. Include any authors and reviewers required.*