This has got to be an error in the examiners' report. Part (iii) calculates the standard formula total SCR as 563.7. I've checked ASET and they use the correct figure.
Looking at the standard formula first:
Undiversified insurance risk is 199.4+384.5=583.9.
Diversfied insurance risk is 475.3 (from part (iii). The difference is 108.6, or 18.6% of the total.
Comparing to the internal model:
Undiversified insurance risk is 74.5+59.5+389.9=523.9.
Diversification benefit for insurance risk is 158.7, or 30.3% of the total.
Mind you this might be expected (to an extent) because the internal model diversification will also include an extra element of diversification between premium risk and reserve risk that wasn't allowed for in the QIS5 correlation assumptions.