Standardise data-raw scripts

[hadley]

• Always produce standard tbls (not spec_tbl created by readr)
• Fix deprecated functions
• Standardise package loading
• Create csvs for all external datasets

Fixes #1459

---

I verified that all datasets had the correct type with

data <- dir("data", full.names = TRUE)
data |> lapply(function(path) {
  env <- new.env()
  name <- load(path, envir = env)
  class(env[[name]])
})

and that all scripts worked in a fresh session with

scripts <- dir("data-raw", pattern = "\\.R$", full.names = TRUE)
for (script in scripts) {
  cat(script, "\n")
  callr::r(function(path) {
    source(path)
  }, list(path = script))
}

[DavisVaughan]

cms_patient_experience seems to have changed a lot in the CSV, is that worrisome? It looks like it pulls from an API

[hadley]

Good point; let's treat the cached csv files as the source of truth.