move `raw-data/` tests inside of `raw-data/` script

[zkamvar]

in #33, a test suite was added to test the validity of the raw data. This is a good thing! The icky thing is the fact that the package tests were never really set up for this situation, which leads to things like getwd() sneaking into the code.

Given that these tests are designed to test the data generation and will only ever run locally, would it make more sense to move them inside of the script?

library(testthat)
library(dplyr)

## Data fetching and generation

test_that("data looks good", {
  expect_is(...)
})

test_that("data smells good", {
  expect_equal(...)
})

## Data writing and stuff.

[annakrystalli]

Just to add that if it is data integrity we are interested in, I.e. checking the data for certain characteristics before saving) then the assertr package might be more appropriate. https://docs.ropensci.org/assertr/

[zkamvar]

Oh yeah! assertr is fantastic. I've used it before in my analyses. I really like the feature where you can chain the assertions so that if you have multiple sources of failure, it will report all of them instead of just failing on the first one: https://docs.ropensci.org/assertr/articles/assertr.html?q=chain#combining-chains-of-assertions