List of things I know I need to do:
-----------------------------------

Store the id of the sect with the title of the graphic so that I can get
xref correctly - (Ali: I looked at this item and there seems no point in doing
this. We can only examine graphics in the /CURRENT/ sect-id (the one specified
on the command-line) using the pre-parser. This means we cannot properly xref
figure's outside the current sect)

Store sect level so we can do <H2>/<H3> correctly for section headers - This
appears to be done (if this is meant for the table of contents) 

Implement variablelist

Switch to a DOM based parser.
