filter_datasets was by default using the "only valid" datasets logic, which in turn prepped the datasets for use in likelihoods and called the expensive MF sampling function.
Considering this is only ever used in the plotting to determine what datasets exist (other places call filter_likelihoods instead), there is no need for the more expensive operation to be called by default.
filter_datasetswas by default using the "only valid" datasets logic, which in turn prepped the datasets for use in likelihoods and called the expensive MF sampling function. Considering this is only ever used in the plotting to determine what datasets exist (other places callfilter_likelihoodsinstead), there is no need for the more expensive operation to be called by default.