Rlmodels

- Added progress bars
- Fixed bug in DDPG by casting actions to numpy arrays
- updated parameters and gym environment names in example scripts

Modified network class name

also:
- bugs fixed in n-steps TD handling
- error messages added when necessary
- now the episodic reward is plotted instead of average timestep reward
- docs improved

also:
- logging improved
- experience replay improved

- DDPG added
- logging added

- scheduers added to models for runtime hyperparameter tunning
- implemented importance sampling in DQN
- improved specification of weight function in CMAES