OfficeImporter does not import word documents containing internal frames correctly

This is because of two reasons:

1. OO server generates invalid xhtml for this kind of documents where it puts paragraph elements inside span elements (which is illegal xhtml)
2. Even if it generates div elements instead of span elements (correct way) our rendering module does not generate embedded documents for such constructs at the moment, see http://jira.xwiki.org/jira/browse/XWIKI-3262?focusedCommentId=38392&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#action_38392

Both of these issues can be easily worked around in officeimporter html cleaner.