Investigate reading from content.xml

[jesse-gallagher]

A p2 repo's content.xml contains most (all?) of the metadata that we end up gleaning from the Jars themselves, but also tends to dwarf the size of the artifacts.xml file.

It would mostly be useful for #19 , in order to generate an internal dependency tree based on Import-Package, if that ends up being desirable.