R dplyr using across() efficiently with mutate() and case_when()

Question

Here is data:

library(tidyverse)
set.seed(1234)
df1 <- tibble(
  S1 = rnorm(5), 
  S2 = rnorm(5), 
  S3 = rnorm(5), 
  S4 = rnorm(5), 
)

I