Tidycensus

* Given that the Census API allows for 500 queries per day without an API key, the API key requirement in the package has been removed to support reproducibility. Users without a key are now warned of potential performance limitations.
* PUMS variables for 2021 have been added to the `pums_variables` data dictionary object.
* Various bug fixes and performance improvements; browse [the GitHub issues](https://github.com/walkerke/tidycensus/issues) for a complete listing.

* Minor release with bug fixes and performance improvements. Fixes include support for the 2000 SF3 Data Profile; improvements to the `interpolate_pw()` function; and correct geometry usage for PUMA geography in the 2021 ACS.

* `get_acs()` and `get_pums()` now default to `year = 2020` to retrieve data from the 2016-2020 5-year ACS. 2020 1-year data are not available in tidycensus and the package throws an error message; users should hard-code a different year to get 1-year ACS data.
* A new `as_dot_density()` function allows users to quickly generate dots from data for dot-density mapping. This function requires the terra package to be installed.
* A new `interpolate_pw()` function supports population-weighted interpolation of data between incongruent geometries, based on Esri's weighted block centroid apportionment algorithm. A common use-case will be interpolating pre-2020 data to 2020 geometries like Census tracts and block groups.
* `get_pums()` now correctly returns vacant housing units either when the `VACS` variable is requested or when `return_vacant = TRUE`.
* `get_acs()` now supports the ACS Comparison Profile data, helping users perform longitudinal demographic analyses correctly.
* `load_variables()` now returns a `geography` column for the 5-year ACS Detailed Tables that tells users the smallest geography at which any given variable is available.
* Numerous bug fixes and performance improvements.

* tidycensus now supports the 2020 decennial Census PL-94171 redistricting data. Use `year = 2020` in a call to `get_decennial()` to access the data, and `load_variables(2020, "pl")` to look up variable IDs.
* A new geography, `"voting district"` is now available for 2020 to assist with redistricting analyses.

* Support for the ACS Migration Flows API is available in a new function, `get_flows()`. [Browse the supporting documentation](https://walker-data.com/tidycensus/articles/other-datasets.html#migration-flows-1) for information on how to analyze and map migration flows data in tidycensus.
* The `shift_geo` argument is deprecated in favor of using the new `tigris::shift_geometry()` function. This allows for shifting of geometries for any geography in national mapping projects.
* Various bug fixes and performance improvements; browse [the GitHub issues](https://github.com/walkerke/tidycensus/issues) for a complete listing.

* Added support for the 2015-2019 5-year ACS PUMS data in `get_pums()`. Updated variable definitions are found in the data object `pums_variables`.
* The default year in `get_pums()` is updated to 2019.
* For 2015-2019 ACS aggregate data, `get_acs()` now supports the `state` parameter for zip code tabulation areas (ZCTAs) to allow subsetting by state.
* Various bug fixes and internal improvements; browse [the GitHub issues](https://github.com/walkerke/tidycensus/issues) for a complete listing.
* Added a `NEWS.md` file to track changes to the package.