promote disorder/secondary structure predictions to dataset annotation

Currently, sequence based predictions made by the JABA disorder predictors and JPred are added to the sequence held in the visible alignment, rather than its dataset reference sequence. In JPred's case, this fits with the rationale that the returned profile was used to generate the JPred prediction, but in JABA disorder predictors, the result is solely based on the service implementation and input data.

Ideally, predictions should be added to the dataset where appropriate to allow them to be propagated in the same way as features, and accessible via the 'Add reference annotation' function.