relax SVN archive

Added the relaxation dispersion dataset recorded at 800 MHz in the system-test.

This was kindly provided by Dr Flemming Hansen (flemming AT pound DOT med DOT utoronto DOT ca) and
was previously published in: Hansen, Vallurupalli & Kay, 2008, JPhysChemB, 112: 5898-5904.

The original format was different and two formats were made ('generic' and 'sparky'), as for the
dataset recorded at 500 MHz.

Created a directory for the data recorded at 800 MHz and put a readme file explaining its origin.

The readme file was copied using the command:
svn cp test_suite/shared_data/curve_fitting_disp/dataset_1-a/readme test_suite/shared_data/curve_fitting_disp/dataset_1-b/readme
and was subsequently modified.

Added details to the readme file and changed the directory name where the sample data is located.

The directory is now named 'dataset_1-a'. This contains data recorded at 500 MHz. Data recorded at
800 MHz will be put in a directory called 'dataset_1-b'.

Wrote a basic model-free bmrb_read() method.

This simply parses the STAR formatted file, but does not yet put the contents into the relax data
store.

Fix for the write() function. The directory for placing the file into is now created.

Fixes for the read() function.

Fix for the test_rw_bmrb_model_free() system test.

The dir name was being set to the data pipe name!

The pystarlib File object is now initialised and the write() method called.

This produces a relatively empty file.

Added a printout for the write() function.

Redesigned the generic BMRB write() function for the pystarlib.File.File class object.

The open_write_file() function should not be used as pystarlib already creates the file.