Use proper tidyselect tidyselections in epi_slide_opt, or don't call tidyselection

[brookslogan]

We currently use a limited form of tidyselection, so something like this goes awry in a weird way:

library(epiprocess)
#> 
#> Attaching package: 'epiprocess'
#> The following object is masked from 'package:stats':
#> 
#>     filter
jhu_csse_daily_subset %>% group_by(geo_value) %>% epi_slide_mean(before = 6L, where(is.numeric))
#> Error in `.data_group[, col_names_chr]` at epiprocess/R/slide.R:751:7:
#> ! Can't subset columns that don't exist.
#> ✖ Columns `where` and `is.numeric` don't exist.
#> Backtrace:
#>      ▆
#>   1. ├─jhu_csse_daily_subset %>% group_by(geo_value) %>% ...
#>   2. ├─epiprocess::epi_slide_mean(., before = 6L, where(is.numeric))
#>   3. │ └─epiprocess::epi_slide_opt(...) at epiprocess/R/slide.R:863:3
#>   4. │   └─mutate(x, .real = TRUE) %>% ... at epiprocess/R/slide.R:776:3
#>   5. ├─dplyr::group_modify(., slide_one_grp, ..., .keep = FALSE)
#>   6. ├─epiprocess:::group_modify.epi_df(., slide_one_grp, ..., .keep = FALSE) at dplyr/R/group-map.R:156:3
#>   7. │ └─dplyr::dplyr_reconstruct(NextMethod(), .data) at epiprocess/R/methods-epi_df.R:231:3
#>   8. │   └─dplyr:::dplyr_new_data_frame(data) at dplyr/R/generics.R:196:3
#>   9. │     ├─row.names %||% .row_names_info(x, type = 0L) at dplyr/R/utils.R:18:3
#>  10. │     └─base::.row_names_info(x, type = 0L) at dplyr/R/utils.R:18:3
#>  11. ├─base::NextMethod() at epiprocess/R/methods-epi_df.R:231:3
#>  12. ├─dplyr:::group_modify.grouped_df(., slide_one_grp, ..., .keep = FALSE)
#>  13. │ ├─dplyr::group_map(.data, fun, .keep = .keep) at dplyr/R/group-map.R:193:3
#>  14. │ └─dplyr:::group_map.data.frame(.data, fun, .keep = .keep) at dplyr/R/group-map.R:124:3
#>  15. │   └─dplyr:::map2(chunks, group_keys, .f, ...) at dplyr/R/group-map.R:145:5
#>  16. │     └─base::mapply(.f, .x, .y, MoreArgs = list(...), SIMPLIFY = FALSE) at dplyr/R/import-standalone-purrr.R:66:3
#>  17. │       └─dplyr (local) `<fn>`(dots[[1L]][[1L]], dots[[2L]][[1L]])
#>  18. │         └─epiprocess (local) .f(.x, .y, ...) at dplyr/R/group-map.R:180:5
#>  19. │           ├─data.table (local) f(...) at epiprocess/R/slide.R:751:7
#>  20. │           │ └─data.table:::froll(...)
#>  21. │           ├─.data_group[, col_names_chr] at epiprocess/R/slide.R:751:7
#>  22. │           └─tibble:::`[.tbl_df`(.data_group, , col_names_chr) at epiprocess/R/slide.R:751:7
#>  23. │             └─tibble:::vectbl_as_col_location(...)
#>  24. │               ├─tibble:::subclass_col_index_errors(...)
#>  25. │               │ └─base::withCallingHandlers(...)
#>  26. │               └─vctrs::vec_as_location(j, n, names, missing = "error", call = call)
#>  27. └─vctrs (local) `<fn>`() at vctrs/R/subscript-loc.R:80:3
#>  28.   └─vctrs:::stop_subscript_oob(...)
#>  29.     └─vctrs:::stop_subscript(...) at vctrs/R/subscript-loc.R:481:3
#>  30.       └─rlang::abort(...) at vctrs/R/subscript.R:144:3

^{Created on 2024-04-09 with reprex v2.0.2}

We should consider proper {tidyselect}-based tidyselections, probably based on tidyselect::eval_select(<look at docs>, allow_rename = FALSE) unless we want to allow "renaming" to do something, and either

• use them, update roxygen to follow tidyverse/roxygen's way of marking tidyselection args (example here), [update arg forwarding from epi_slide_mean if needed -- {{ }} might be needed] add a test using tidyselect features --- maybe based on the above failing example; or
• don't call what we do tidyselection.