Data structure changed so that each dds contains its own model. This means that, if applying a time variable effect (eg. primary beam) to a static model, the invariance of the model needs to be baked into the prior.
First go at implementing a distributed deconvolution in spotless worker