The crowd-persistence module needs to split out the HIbernate implementation details from the shared InternalDirectoy needed by apps that embed Crowd.

Apps that embed Crowd need the InternalDirectory classes that are here, but it also brings in other dependencies that are not desired.

eg:

• Crowd's Hibernate DAO implementations (eg UserDAOHibernate)
• org.hibernate transitive dependency
• atlassian-config (which is incompatible with DBCP as used by JIRA and seems to be Hibernate specific).