relax SVN archive

Updated the release checklist.

Updated the release checklist file.

relax version 1.2.0.

This is the first release of the new 1.2 branch. Relaxation curve fitting is now supported within
relax. Many of the development related scripts for compiling the manual, removing temporary files,
creating the distribution packages, etc have been moved into makefiles. Installation is now through
makefiles as well and the command 'make install' as root will install relax. C modules have been
added which interface with relax through the Python/C API. These implement the functions using in
the optimisation of relaxation curves. To create the C shared objects, the command 'make' is used.

Creation of the 1.2 stable branch.

Many of the steps in releasing stable relax versions have been shifted into makefiles.

The makefile has had two rules added, 'binary_dist' and 'source_dist'. The 'binary_dist' rule will
create the distribution archive files by first building the C modules, compiling the manual,
removing the temporary *.pyc, *.bak, and *.o files, archiving the files of the base directory
(excluding .svn directories) into the file relax-x.x.x.os.mach.tar.bz2 where x.x.x is the release
number taken from 'version.py', os is the name of the operating system, and mach is the machine the
C code was compiled on. Finally the file will be GPG signed (requires the relax private key!). The
'source_dist' rule will do the same but without the C compilation and all *.so files will be removed
as well.

The file 'docs/release_checklist' has been added. The file describes all the steps required for
releasing new relax versions.

The makefile 'docs/latex/Makefile' has been improved and a rule added for creating the manual
without executing relax.

The TODO file has been updated. Relaxation curve fitting has been removed as the code is complete.
It now has its own section describing the missing features of the technique.

The Swig file 'maths_fns/swig_jw_mf.i' and corresponding C file 'maths_fns/swig_jw_mf.c' have been
removed as they are not used and Swig was abandoned a while ago.

Added the checklist file describing the steps required for releasing stable versions of relax.

Merge of r2329 from 1.0 branch

Shifted the program version number to its own file so it can be read by the Makefile.

Relaxation curve fitting and error analysis is now fully functional!

The ability to execute Monte Carlo simulations has been added, as well as Grace plotting
capabilities. Now all parts of relaxation curve fitting data analysis are functional. However, the
'dfunc()' and 'd2func()' C functions have not be implemented and therefore only simplex minimisation
without constraints is supported.

The standard deviation in the relaxation curve fitting C module is now a Python array rather than
a double. The data structure called 'intensities' has been changed to 'values' so that the
'maths_fns/c_chi2.c' chi-squared function is made more generic.

In the common functions of 'specific_fns/base_class.py', the 'set_error()', 'sim_init_values()',
'sim_return_param()', and 'sim_return_selected()' functions have been added for Monte Carlo
simulation functionality. The function 'sim_pack_data()' which was moved from
'specific_fns/model_free.py' to 'specific_fns/base_class.py' has been returned as it is not
generalisable.

A few bugs have been removed from 'specific_fns/relax_fit.py'. In the function
'assemble_param_vector()', Monte Carlo simulations are now properly supported. The
'create_mc_data()' function has also been improved. The function 'set_error()' has been removed
from the file. This function was a relic as the origin of 'specific_fns/relax_fit.py' was from
'specific_fns/noe.py'. It hides the function of the same name in 'specific_fns/base_class.py' and
hence causes Monte Carlo simulations to fail.

The functions 'read_columnar_results()', 'write()', 'write_columnar_line()', and
'write_columnar_results()' have been deleted out of 'specific_fns/relax_fit.py'. These too were
relics from the file's origin from 'specific_fns/noe.py'.

The calculation of the peak intensity errors has been modified. The docstring in
'prompt/relax_fit.py' has been rewritten and significantly improved. Now if all spectra have been
replicated, each relaxation time point will have its own error (which is averaged across all
residues). If there are time points with only a single spectrum, then the standard deviations are
averaged for all time points.

In the data analysis chapter of the manual, the section on relaxation curve fitting has been half
completed.

Fix the noe_ri_table bug introduced in r2324