update Getting_Starting_Training.md to reflect current code in training_navigation.py

Made some small changes to Getting_Started_Training.md to reflect the current code in training_navigation.py and flatland, and added some explanatory text in places.

As some people may prefer work directly from the the tutorial text I added some of the code necessary for training and tested locally that Getting_Started_Training.md includes all the code necessary to get the training loop to run. That last change meant the addition of a slightly bulky code block containing all the needed logging and training parameters, but I'm open to comments on whether that makes the whole thing too verbose and that particular code chunk should remain in training_navigation.py instead.