Use `survey` package to estimate TATE and se in weighting()

[benjamin-ackerman]

I think survey estimates a model-robust standard error unlike the lm package, would suggest using svydesign() and svyglm() to estimate the TATE when using weighting methods