Better documentation

[gdalle]

We urgently need a better documentation with

• a better intro to the theory
• examples for all types of loss functions
• examples for using kwargs to pass instances (Clarify & demonstrate use of kwargs #56)
• a guide for choosing your layer à la scikit-learn

[gdalle]

I also wanna write an explanation of the design of the library:

• there are layers and losses, all callable
• some layers are probabilistic, others not, but all are differentiable
• most losses must be built with a layer
• if you have your own differentiable loss you can compose it with a layer
• if you have your own non-differentiable loss you can pushforward it with a probabilistic layer