Removed the extra hash from the header text by counting correctly.
Implemented lists! Both unordered and ordered lists have been
implemented, and they can even be mixed.
Lists starting with letters or roman numerals is not implemented, and
probably won't be.
Added a bunch of code to free created objects. One notable change was
to return the next token from FreeToken() to cut down on boiler plate
code that kept track of a "consumed" token. Now all that just happens
in FreeToken() and I don't have to worry about that anywhere else.
Also added FreeNode() to free all nodes. This wasn't done before at
all. Like FreeToken(), it will return the next node in the list to
avoid having to track that stuff in the calling context.
This node handles both plain paragraphs as well as block quotes. Single
newlines are turned into spaces, double newlines terminate the block.
Care is taken to not have repeated whitespace tokens in the content,
while keeping the number of spaces.