relax SVN archive

Massive speed up of the bond_vectors() method.

The algorithm used was very, very poorly designed. Now when setting the profile flag to 1 gives the
following. Before in one test:

11168736 function calls (10986641 primitive calls) in 76.750 CPU seconds

ncalls tottime percall cumtime percall filename:lineno(function)
758954 12.160 0.000 51.060 0.000 /home/edau/relax/relax-1.3/generic_fns/relax_re.py:31(search)
141 0.890 0.006 67.620 0.480 /home/edau/relax/relax-1.3/generic_fns/structure/internal.py:444(bond_vectors)

All other function call info lines have been removed. After the changes:

2586419 function calls (2579305 primitive calls) in 21.370 CPU seconds

ncalls tottime percall cumtime percall filename:lineno(function)
62803 1.180 0.000 5.020 0.000 /home/edau/relax/relax-1.3/generic_fns/relax_re.py:31(search)
141 0.060 0.000 12.020 0.085 /home/edau/relax/relax-1.3/generic_fns/structure/internal.py:444(bond_vectors)

The difference is that the selection object is no longer used, as it is completely unnecessary.

Modified 2 system tests for the loading of N-Ca vectors.

Modified the get_pipe() function to check if a data pipe exists.

Modified a print out to make it clearer.

Added a check for the molecule name.

Bug fix for the atom_loop() method.

The molecule name was not being checked!

Modified test_read_pdb_internal1() to catch a molecule name identification bug.

The test_read_pdb_scientific1() was similarly modified, but there is no bug in the Scientific python
code.

Fix for the test_read_pdb_complex_scientific() system test.

Retarded Scientific python splitting a perfectly good molecule into 2 when it can't recognise a
residue type!

Fixed an import (as well as a few comments).

This however introduces an error concerning the 'chi2' being undefined in the C module for
relaxation dispersion...

The read_mol arg now does something in load_pdb()!!!