Make indexer not dependent on XPath

[Kit Blake]

An indexer reads all the elements marked as index items in a publication
and creates a web version of an index (like in the back of a book). This
feature is implemented, partially using XPath, but since this introduces
another dependency, we'd like to do it in pure Python, and then make it
available to all Silva users.

[Martijn Faassen (faassen)]

As a first step I checked it into CVS today. Removing the dependency
shouldn't be a big problem.

[Martijn Faassen (faassen)]

The plan is to include the indexer in the Silva core, right?
As soon as it's not dependent on XPathIndex anymore it's only
dependent on PyXML, which is already a requirement for Silva
through ParsedXML.

[Kit Blake (kitblake)]

This is a terrific feature, and we should include a demo of
the indexer in the demo for the next release. Let's not
resolve this issue until the demo is done.

[Martijn Faassen (faassen)]

Okay, now it should be added to the Silva CVS. Its preview
breaks on a default install, but I think that's because we
don't automatically install the page templates yet. If I
fake the existence of a 'frontend.css' it works.

One limitation is that it doesn't index 'index' documents
yet.

[Martijn Faassen (faassen)]

This seems to be working, and is thus resolved.