Remove spec() from dataset generation scripts

[olivroy]

Summary

Fix #1697

The next time datasets are built, they will remove the readr::spec() attribute. Other datasets don't have this issue.

This is very minor, but it is just that object comparisons will be a little bit easier if we drop these classes.

I didn't rebuild the data as this is minor and this will naturally occur next time you need to add a dataset.

Strategy found in readr 1.3.0 changelog https://readr.tidyverse.org/news/index.html#tibble-data-frame-subclass-1-3-0

Basically, this aims to solve this.

v1 <- gtcars %>% dplyr::slice_head(n = 1, by = year) %>% dplyr::arrange(year)
v2 <- gtcars %>% dplyr::group_by(year) %>%  dplyr::slice_head(n = 1) %>% dplyr::ungroup()
all.equal(v1, v2)
#> [1] "Attributes: < Length mismatch: comparison on first 2 components >"                   
#> [2] "Attributes: < Component “class”: Lengths (4, 3) differ (string compare on first 3) >"
#> [3] "Attributes: < Component “class”: 3 string mismatches >"      

WIth the help of this PR, eventually when the datasets are redone,

# gtcars will eventually become gtcars2 with this PR.
gtcars2 <- gtcars[]
v3 <- gtcars2 %>% dplyr::slice_head(n = 1, by = year) %>% dplyr::arrange(year)
v4 <- gtcars2 %>% dplyr::group_by(year) %>%  dplyr::slice_head(n = 1) %>% dplyr::ungroup()
all.equal(v3, v4)
#> TRUE

Note that gtcars2 still has tbl_df class ("tbl_df" "tbl" "data.frame")

[rich-iannone]

LGTM!