Better describe/visualize data levels representing spectrum of raw to processed data

[anngvu]

Currently, templates for data types are broken down into numbered levels. This can be unfriendly because of the numbers being rather abstract, and for most researchers it can be hard to remember the difference between Level 1 and, say, Level 3. One idea to augment the docs is to use one of the ENCODE graphs and draw boundaries around files and their correct data level template.