relax SVN archive

Small change to systemtest Relax_disp.test_r1rho_kjaergaard.

The outcome of kex from systemtest is in the area of 4-5000. The expected value is 13000.
A deeer analysis of the input is needed, to judge what is the correct value.

Modified systemtest Relax_disp.test_r1rho_kjaergaard to use input guess values of phi_ex in units of ppm^2 instead of rad^2/s^2.

Removed links for many old relax releases.

This is for the webpage http://www.nmr-relax.com/download.html and the links for the old releases
now all point to http://www.nmr-relax.com/download_request.html.

Replaced how a global analysis average results from a previous run to instead take the median.

This is to prevent averaging extreme outliers, and instead take the median of the previous result.
This was discussed in: https://mail.gna.org/public/relax-devel/2013-10/msg00009.html.

Changed a unit test and systemtest, where a change from the atan to atan2 function to calculate theta can give differences to the 15 decimal.

Updated the news announcement links for relax version 3.1.7 at http://www.nmr-relax.com/download.html.

The correct link is https://gna.org/forum/forum.php?forum_id=2448.

Updated the http://www.nmr-relax.com/download.html webpage for relax 3.1.7.

Updated the relax version number to 3.1.7 on the http://www.nmr-relax.com/api/index.html webpage.

Updated the relax API documentation at http://www.nmr-relax.com/api/3.1/ to version 3.1.7.

Updated the relax HTML manual at http://www.nmr-relax.com/manual/index.html to version 3.1.7.