I see your point but the reason for the rounding is because we are generally dealing with fairly small cohorts (certainly fewer than 100 in the vast majority of school cases), which of course means that each pupil is worth more than 1% (usually 2-3%). We don't feel that decimals would be meaningful for reporting in such cases where cohorts are small. It does of course result in rounding errors, which is inevitable, but we felt that integers were clearer. It's the same approach taken by the DfE in the performance tables where KS2 results are reported as whole percentages.
You mention that the Insight rounds up to the nearest integer, It should be rounding up and down depending on which side of 0.5% it falls (>=0.5% rounded up). Can you check that and let us know.