I've noticed that some metastable nuclides (such as 471080001 and 330820001) in the feed inventory of the separations archetype that are not placed in any product streams end up not going to the leftover stream, and thus disappear from the system altogether. Note that the decay control parameter of the simulation is set to never.
The disappearance of 471080001 (aka Ag-108m) is particularly odd since its half-life is > 400 years.