Problem using previous samples to new inference in new model context

[ricardoV94]

Alternatively, the standard and safer approach is to just resample the whole model with the bigger dataset, instead of doing it sequentially. Sometimes this is not even that much slower