relax SVN archive

Pre-defined probabilities can now be used in the 'fixed' N-state model optimisation.

Fixed the align_tensor.delete() arg unit test to accept None as an arg.

The align_tensor.delete() user function can now be used to remove all tensors simultaneously.

Made a RelaxError less stringent so that the paramagnetic centre can be unfixed.

Added a system test to back-calculated RDCs from a PCS derived alignment tensor.

The RDC and PCS back-calculation user fns now add the alignment ID to the RDC and PCS if absent.

Again this is to allow both data types to be back-calculated in the absence of measured data.

Small fixes and improvements to the rdc.back_calc() user function.

Align IDs are being used instead of RDC IDs so RDCs can be back-calculated even if measured RDCs are
not present. And spin info is also being checked for better error reporting to the user.

The Ln3+ position optimisation system test has been sped up by decreasing the accuracy.

Fix for the alignment tensor initialisation for when the ID is added to cdp.align_ids.

This was introduced in r11301. cdp.align_ids was not being checked if the align ID already exists.

Added some unit tests for the pseudo-elliptic cosine function.