Split the giant monolith sample_joint into parts, to make code reuse easier for custom posteriors e.g. in pipeline B2. Also slightly changes how sampling is done, intercommunication now happens after every step which simplifies code a bit (performance seems ok), and we print out a TimerOutputs.jl table as well.
Split the giant monolith
sample_joint
into parts, to make code reuse easier for custom posteriors e.g. in pipeline B2. Also slightly changes how sampling is done, intercommunication now happens after every step which simplifies code a bit (performance seems ok), and we print out a TimerOutputs.jl table as well.