Supervised Learning Revamped and Stable

Rewrote much of supervised learning. Added explanations of optimization, loss functions, activation functions. Added missing (binary) logistic regression section.

Dogfood -> Alpha

This release marks that every section has received at least one solid preliminary editorial pass/rewrite. Introduction and much of probability have received rewrites, fine-tune and object detection, and optimization now have introductions. Since the last release we also have added considerable content, including a draft of a DQN, GANs, optimization, etc.

Primordial book draft

Initial draft seed of a book on deep learning, in theory, in application, and in MXNet.