relax SVN archive

Removed a debugging print out from generate_sequence().

The reading of BMRB model-free saveframes will now generate sequence data if needed.

Shifted the sequence generation code into the BMRB specific generate_sequence() function.

This is to allow for strangeness in the BMRB entries :S

Reverted r12461 as this was the wrong way to do this.

The command used was:
svn merge -r12461:12460 .

The number of spins is now constructed from the 's2' var if 'data_ids' does not exist.

This allows v2.1 model-free saveframes to be read.

Removed the molecule name generation from the entity data, as the entity data is not always there.

This fixes a crash when reading BMRB files with no entity saveframe.

Spin names are not used in the spin ID strings as the AtomName tag is not compulsory.

Fixes for 2 model-free system tests for a new system which again shows optimisation differences :S

The results are from Tiago Pais <tpais att itqb dott unl dott pt> (Message-id:
<BA08C06336CE4466990EBAFB59D7DFCF@...>,
https://mail.gna.org/public/relax-users/2011-01/msg00029.html.

Updated the API documentation (http://www.nmr-relax.com/api/1.3/) for relax 1.3.9.

Updated the API doc version number to 1.3.9 (http://www.nmr-relax.com/api/).