relax SVN archive

Changed the way optimisation of the N-state model is set up.

This will allow multiple base data types to be used simultaneously.

Added a module docstring.

Fixed all the calls to __base_data_types() and subsequent use of the returned list.

Fixed __base_data_types() to properly identify tensors as the base data source.

Modified the __update_model() to handle both RDCs and PCSs.

Redesigned __base_data_types() to determine all base data types, and return them as a list.

Renamed __determine_data_type() to __base_data_type().

Created 2 new RelaxErrors for PCSs.

Converted the read() function to handle PCS rather than RDC data.

Changed a lot of references from cdp.rdc_ids to cdp.align_tensors.