relax SVN archive

Fix for bug #14885 (https://gna.org/bugs/?14885).

This was reported by Tiago Pais (tpais att itqb dot unl dot pt).

The problem was a few missing user functions in the sample script.

Improved the warning messages for over-fit spin deselection.

Missing 'warnings' import.

Modified all the specific analysis overfit_deselect() methods to warn when spins are deselected.

Added the back end to the rdc.back_calc() user function!

Split up relax_gui.py to make it smaller.

This change was communicated by Michael Bieri (https://gna.org/users/michaelbieri) in the posts
https://mail.gna.org/public/relax-devel/2009-12/msg00020.html (Message-id: <4B217970.805@...>) and
https://mail.gna.org/public/relax-devel/2009-12/msg00022.html (Message-id: <4B21950C.5080803@...>).

This enhances clarity allowing for a better understanding and to facilitate manipulation. The
relax_gui.py module was split up as follows:

relax_gui.py
--> res/calc_modelfree.py (script to calculate model-free).
--> res/calc_noe.py (script to calculate noe).
--> res/calc_rx.py (script to calculate Rx).
--> res/project.py (script to open and save the project)
--> res/results_analysis.py (script to perform post-analysis: craete grace plots, pymol macros...)

The res/filedialog.py, res/message.py, res/select_model_calc.py, and res/settings.py have not yet
been created due to ambiguities as to their origin.

The temperature value is now converted into a float before storage.

Big bug fix for the RDC and PCS calculations used in the fixed population N-state models.

The last weight, if supplied, was being converted to zero (1 - sum(weights)) hence the last
state/structure was not being included and the population sum was not 1!

Fix for the PCS back calculation for when no weights are given.

Added a check to test_R_to_euler_zyz() for beta angles > pi.