Support added in r1331618. We can now get headers, footers and footnotes, assuming a file only has one set of each, with the default names. (If a file has multiple styles with different ones, the code will likely just end up with the last one)
Note that we are rapidly approaching the point when the current model for the parser won't cope. At that point, we'll need to start holding things like styles, headers, footers etc properly, track state more as we process the file (a single state level isn't really enough), be aware of styles applied to text etc.