Replace utils read with readr read function

[r-ash]

Was just having a quick look at anc_spectrum_warning and art_spectrum_warning as we will use similar code to build the data for the new input plots. The majority of the time running these functions is just reading the spectrum files. Particularly for large countries. I've replaced the reading with utils::read.csv with readr as it is quite a bit faster.

For DRC, read_dp is the function before this change, read_dp2 is the function after this change. Note the only difference should be is that where read_dp would return columns with the types set by R. They are always "character" now. I believe this is fine because we have to manually convert any data later anyway (see e.g. read_dp_art_dec31) as the type interpretation is often broken due to how DP files are stored in disk.

bench::mark(read_dp(data$pjnz), read_dp2(data$pjnz), check = FALSE, iterations = 5)
# A tibble: 3 × 13
  expression             min median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result memory     time       gc      
  <bch:expr>           <bch> <bch:>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list> <list>     <list>     <list>  
1 read_dp(data$pjnz)   313ms  341ms      2.44    75.5MB     1.95     5     4      2.05s <NULL> <Rprofmem> <bench_tm> <tibble>
2 read_dp2(data$pjnz)  176ms  188ms      5.31    24.1MB     1.06     5     1   941.76ms <NULL> <Rprofmem> <bench_tm> <tibble>

[jeffeaton]

Nice spot. Thanks. This is something that might have benefit for changing across other packages used here (EPP, Shiny90, leapfrog, etc.)

Bigger rewrites could be done to avoid re-reading the DP file multiple times.