Develop import/export XML schemata

[rafaelnajera]

The current schema does not reflect the way documents, transcriptions and chunks are handled in the system. Besides, the system can generate different kinds of object:

• Full transcription of a subset of pages of a document
• Full transcription of a chunk in a particular witness
• (Eventually) Critical edition of a collection of chunks

For full transcriptions, the following should be included:

• new item types: math text, paragraph mark, etc
• simpler handling of chunk markers
• system ids for documents (instead of DARE codes)
• proper, simplified TEI header

[rafaelnajera]

Not really needed for the time being