`remove.dups` in `strech()` suffers from performance issue

[psychelzh]

Reproducible example

When add remove.dups = TRUE, stretch() costs too much time:

x <- matrix(rnorm(100 * 206), nrow = 100) |> as.data.frame()
cors <- corrr::correlate(x, quiet = TRUE)
system.time(corrr::stretch(cors))
#>    user  system elapsed 
#>    0.02    0.00    0.08

system.time(corrr::stretch(cors, remove.dups = TRUE))
#>    user  system elapsed 
#>    1.79    0.19    6.72

^{Created on 2024-09-13 with reprex v2.1.0}