Introduce identifiers as a domain entity for people entities

In the current data model, Students and Staff have a unique ID and then an array of other identifiers (as a subclass on a StudentEdOrgAssociation or Staff). This structure has proven to be non-optimal for a few reasons:

• It does not allow for identity to be managed easily in contexts where multiple systems are interacting. For example, a system cannot easily write its own identifiers without being given access to one of these entities.
• Uniqueness is hard to enforce and non-uniqueness detect, as values are  held on subsidiary arrays.
• The location of student identifiers on StudentEdOrgAssociaton is misleading, as many identifiers are not dependent on local membership (though it works for some IDs)
• The array of other identifiers metadata is weak and does not assume a rich ecosystem of providers - for example, the identifying system should probably be a namespace and not a descriptor, to reflect that provenance may be unknown.