Good posts Genghis! Yet I am still uneasy that a key point is being missed somewhere. We know from CAP 403 that risk assessment uses the formula Risk = severity of risk x likelihood of occurrence.
Severity, in the event of a crash, runs from potential minor injury (score 1) to multiple deaths (5). Likelihood runs from highly unlikely (1) through possible (3) to highly likely (5). So using CAA's example scenario where a crash could cause 'bystander' casualties, you might have severity at 5 x likelihood at 1 = 5. But if you close a road where non-paying guests might be gathered, you may mitigate the risk down to 2 (very low risk).
Yet, although the pilot's DA must have the box ticked recommending 'loops and barrel rolls in ex-military aircraft' signed off by the examiner, the risk assessment is benign here. I argue that the the risk-likelihood of downward vertical manoeuvres in high inertia jets should be rated as at least possible, especially if the entry height is low, and there are plenty of sad examples around the world to back this up.
If the risk cannot be mitigated sufficiently, then remove the manoeuvre from the sequence.