Confluence importer should keep order of the pages from the source package

The Concluence importer is not really fully streamed because it would not be possible with the Concluence package format. So it first prepare another temporary FS (not not take too much memory) format to import and in the process loose the original order of the pages.

Even if I doubt there is any concept of on purpose order in the Confluence package (looks more like a XML dump of the database then anything else) it's still nicer to follow it on our side (at least for test if for anything).