The mavenization effort is a huge amount of work that would be great to get done, but it's probably too big (x3 - thanks project split!)
Its nothing like 3x the work, if done right. Most of the work would end up in a parent project pom.xml(s) that all 3 projects would inherit from. They could then be built together or independently.
For Avro, the dropping of the JDK5 requirement is huge and we may not switch soon without that as a motivating force. But other tools, like Sphynx (http://sphinx.pocoo.org/) are interesting in part because they can have a custom look/feel easily and ... look good We also have to support build artifacts and documentation for many languages so the python origins of something like Sphynx isn't a barrier to entry.
Maven provides a lot out of the box, but it is ugly and the site will look like a maven project unless customized. Its great for what it does by default trivially but is built for convention and not customization. Build a project with Maven and you get a site for free with zero effort and some extras with little work at all. But it is focused on build artifacts and a few faq-like pages. Build a maven plugin and the docs it generates are beautiful right out of the box – coming from annotations /javadoc in your classes. Want to add a bunch of custom documentation or document types? significantly change look/feel? – not so trivial.