relax SVN archive

Fixed the reversion of the many changes to 'specific_fns/model_free.py' and 'errors.py'.

This occurred in the last revision.

Fix for bug #9101 (https://gna.org/bugs/?9101).

The relax_data.read() user function was failing when the data or error column contained the text
'None' which is the default output for deselected spin systems in the exponential curve fitting
component of relax. This is now fixed.

Improvements to the 'full_analysis.py' sample script.

The script's documentation has been shifted from the __init__() method docstring to the module
docstring.

All of the options which need to be modified by the user have been shifted to the top of the script
as global variables. This means that the user doesn't even need to look at the complex code to
operate the script.

Hand ported r3285 from the 1.2 line, the fix for bug #8953 (https://gna.org/bugs/?8953).

.....
Fix for bug #8953 (https://gna.org/bugs/?8953).

The bug was introduced by a fix to the model-free results writing function. In relax versions 1.2.9
and below, no data about a residue was being stored in the results file if it was deselected. This
behaviour undesirable as a spin system could be deselected temporarily and later on reselected. If
this process is deliberately interrupted by using 'results.write()' followed by 'results.read()' to
resume, then there will be data loss.

The fix was simple: check for the presence of the 'data.num_ri' data structure.
.....

Fix for bug #8953 (https://gna.org/bugs/?8953).

The bug was introduced by a fix to the model-free results writing function. In relax versions 1.2.9
and below, no data about a residue was being stored in the results file if it was deselected. This
behaviour undesirable as a spin system could be deselected temporarily and later on reselected. If
this process is deliberately interrupted by using 'results.write()' followed by 'results.read()' to
resume, then there will be data loss.

The fix was simple: check for the presence of the 'data.num_ri' data structure.

More fixes for bug #8996 (https://gna.org/bugs/?8996).

This bug was reported by Carl Diehl.

Although I had, in the previous commit, fixed the non-existant 'data.heteronuc' data structure there
was still an issue because 'data.proton' should have been changed to 'pdb_proton'!

final separation between the multi framework and relax and initial
documentation of multi.processor

Fix for bug #8996 (https://gna.org/bugs/?8996).

This bug was reported by Carl Diehl.

The issue was that the data structure 'data.heteronuc' did not exist and is now called
'pdb_heteronuc'. This issue had been fixed in the 1.3 line but was incomplete in the 1.2 line (the
printing of values was ok but the errors and simulations were incorrect).

Fix for bug #9093.

threaded multi processor support written, multi is now beta!
multi now supports three substrates: mpi4py, uniprocessor, and threads