survtab: error in SE and CI of age-standardised survival

[kseppa]

Non-standardised standard error is correct. Standard error of age-standardised survival should be calculated as follows:
st <- survtab(FUT ~ agegr,breaks = list(FUT = seq(0, 5, 1/12)))
st[,SE.r.e2.as:=sqrt(sum(SE.r.e2^2*w^2)/sum(w)^2),by='surv.int']
over age groups agegr with weights w.

Check standard errors of age-standardised survival measures other than Ederer II, too.

[WetRobot]

Using survtab example, the blue lines show the old confidence intervals and black lines show the new ones. It is not a insignificant difference!

[WetRobot]

So this was closed in f19824b.