For any zone that does not meet the conditions for burning (e.g., density / temperature thresholds), the second iteration of the simplified SDC is essentially a no-op, since the advective term should be the same as from the first iteration. Note: this assumes that we didn't burn in the previous zone either because of the lagged reactive source predictor.
We can detect this and just copy the advective term instead of reconstructing it in these cases. For problems that evolve a bit before we burn (like wdmerger) this could save some time.