relax SVN archive

Bug fix for __find_bonded_atoms().

The loop over the maximum number of connect atoms was failing if the number of 'connected' atoms was
less.

Improvements to the read_spin_data() function.

The data checks now occur during the yield loop. This allows for invalid lines to be skipped while
those with real data are used.

read_spin_data() now gives a warning rather than an error if the sequence data in a line is bad.

Modified the message printed by RelaxInvalidSeqError.

Significant improvements to the __find_bonded_atoms() method.

The maximum number of bonds an element can have is now taken into account so that protons are not
thought to be attached to 2 carbons within a 2 Angstrom radius! The atoms closest to the atom of
interest are now connected. Geometry constraints or other advanced techniques are not yet
implemented.

Modified the relax_data.read() entries in order to reflect the new design.

The relax_data.read() function now absolutely needs arguments for 'res_num_col', 'data_col' and
'error_col'. Hence, this argument is now shown in the sample script in order to avoid the users
lose time figuring out where their script went bad...

Modified the sequence.read() entries in order to reflect the new design.

The sequence.read() function now absolutely needs an argument for res_num_col. Hence, this argument
is now shown in the sample script in order to avoid the users lose time figuring out where their
script went bad...

Removed a debugging print statement.

More fixes for the N-state model for the 'xh_vect' and 'bond_vect' mix up.

Fix for the retrieval of bond vectors for the N-state model.

The object could also be called 'bond_vect' rather than 'xh_vect'.