specific handling for 1:.N --> .I in seq_linter()

[MichaelChirico]

Closes #2101

[AshesITR]

Careful: .I is not seq_len(.N) in the presence of by:

library(data.table)
df <- data.table(grp = rep(c("a", "b"), each = 2L), val = rep(c(1L, 2L), times = 2L))
df[, rx := .I, by = .(grp)][rx > 1, val := 0L]
df
#>    grp val rx
#> 1:   a   1  1
#> 2:   a   0  2
#> 3:   b   0  3
#> 4:   b   0  4
df <- data.table(grp = rep(c("a", "b"), each = 2L), val = rep(c(1L, 2L), times = 2L))
df[, rx := seq_len(.N), by = .(grp)][rx > 1, val := 0L]
df
#>    grp val rx
#> 1:   a   1  1
#> 2:   a   0  2
#> 3:   b   1  1
#> 4:   b   0  2

^{Created on 2023-11-15 with reprex v2.0.2}