It’s not about the amount of measurement points, uncertainties in dollar amounts tend to scale as sqrt. When you error propagate that you see that inflation uncertainty scales with inflation.

Very simple model

Inflation=v1/v2

D(inflation)=sqrt((D(v1)/v2)^2+(v1/v2^2 D(v2))^2)

Suppose

D(v1)=c sqrt(v1)

D(v2)=c sqrt(v2)

D(inflation)=c sqrt(v1/v2^2+v1^2/v2^3)= c sqrt(inflation+inflation^2)/sqrt(v2)

So can be seen that the leading term for uncertainty in inflation is linear with inflation.