An ARR of 0.4% to 0.7% requires a very small chance of contracting the disease - yet a third or a half of the population has.
If the vaccine would prevent 95% of those cases, then ARR would be closer to a third or a half.
That ARR number from the trials was from a short snapshot timeframe, when cases numbers were lower than this Winter’s wave. Resistance to the disease persists longer than the trial period, and continues to reduce risk.
If the virus eventually works its way through the whole unvaccinated population, ARR should approach RRR.
Where's you come up with those wildly different numbers?

Good point. Modeling this stuff requires complex differential equations -- there are many moving parts.
Also, as the virus works its way through the population more people are exposed to the virus reducing the number of hosts for the virus reducing the need for emergency vaccines.