Also update the documentation of supported tokens and features for more clarity.
The program is currently able to lex most useful tokens, and parse constructs associated with them on simple examples. Unit tests are still missing to formally ensure everything's right.