Fix bug in thcovmat construction
Created by: RosalynLP
Since the post data keyword update to the theory covariance matrix there was a bug in constructing the theory covmat. This is because commondata_procs
also collected over group_dataset_inputs_by_process
so you end up with N_proc iterations of the datasets which messes up the indexing of the th covmat down the line.
I've sorted this out by only using the unique datasets from commondata_procs
when constructing dataset_names
. I think this is one of the main issues affecting #991.