Since the default retriever behavior is to store the downloaded files to nbproject/private/cache/retriever, they will
never be committed to a VCS. Thus if someone checks them out and tries to build without having an internet connection,
the build will fail, as external resources are not cached.

We need to re-introduce the project's Populate Catalog action which existed previously and make sure the files it
generates are:
1) properly stored in a folder accessible by version control
2) properly referenced from the project-wide xml catalog