relax SVN archive

Better handling of missing data in the BMRB relaxation saveframes.

The values are now properly scaled and values and errors are inverted to go from Tx to Rx values.

Fixes for pack_data() to allow the errors to be a numpy array.

The relaxation values and errors from the BMRB file are being converted from ms to s now.

Removed a debugging print out.

The optimisation of Monte Carlo simulations no longer changes the Q factors and back calculated data.

There was a bug in which the back calculated RDCs and PCSs, as well as the Q factors were being
calculated for each Monte Carlo simulation but stored in the non-MC sim data structures. Therefore
after MC sims, the Q factor and back calculated data would be that of the last simulation!!!

Added checks showing that Monte Carlo simulations in the N-state model modify the Q factors.

Fix for the calls to RelaxError in the N-state model return_error() method.

Fix for the N-state model error returning method for when errors are not present.

The spin_id variable was not defined.

Fix for the sample label matching - this is only performed if the sample label is given.