rework eval mode; major refactoring

add retro_eval, fix boltzmann spec

L1 fitness norm, code and spec refactor, online eval

PyTorch 1.0, faster NN, Variable Scheduler, working Atari

VizDoom, NN Weight Init, Plotly Update

benchmark; reward scaling

DQN replace, AtariPER

optimize memory

Singleton, CUDA, Distributed

implement PPOSIL, fix continuous action