I've posted a possible solution here. If it is something you're willing to try, then download the tool and the sample files; please follow the document for step by step guidance. If you run into any issues, send me an email using the contact info (support) on the web site.
The cardinality problem, again, is very tricky. The sample I've prepared for you (all the download links are in the document) is one of the test cases I was using, except that I had to come up with a specific template for your file layout. One issue that seems to be subject to debate is how to calculate the value for XPaths that, from an XML Schema perspective, traverse choice compositors. Another, not so controversial maybe, is how to calculate the cardinality for particles under repeating compositors, etc.
