revise the content model to avoid Maven specifics and other problems

The initial content model took some shortcuts for expediency that need to be cleared up:

• the Maven specifics, particularly in the dependency model, need to be moved to extensions (eg, dependencies should only refer to a namespace / project / version / id instead of having classifiers and types which can be added as extended information on the dependency in the same way it is for the artifact itself (then adjust xmlrpc and the metadata creation consumer)
• extras like mailing lists, etc can be moved into a module so the basic model is much more basic
• remove the 0-indexed lists in favour of proper modelling (select a name that can be referred to as well as iterated instead)
• review the metadata document on the wiki for keying ideas to ensure we can always map everything to an URI and document it
• ensure the storage is properly decoupled from the format - in a maven 2 repo, asc/md5/pom need to be able to be mapped onto the metadata, and not as resources themselves, while pom/jar/dll are all resources (even if they might represent the same artifact, particularly in the case of the POM for a given JAR).