Proper handling of sumlevels 250, 252, and 254

[tuchandra]

See censusreporter/censusreporter#34 for an explanation of the problem this solves.

In order for these geographies to be handled correctly, three things are necessary:

• The geographies have to appear in tiger2014.census_name_lookup with information (namely, geoids) that do not contradict the ACS geoheader table.
• The API has to return these geographies.
• Search results and autocomplete have to include these geographies.

This fixes the first of the three parts. This script creates pseudo-tiger tables for sumlevels 250, 252, and 254 so that the geoids correctly correspond to acs2014_5yr.geoheader. It then inserts (replaces if some already exist) rows into tiger2014.census_name_lookup for these sumlevels, so that they appear in API calls, search results, and autosuggest.

[JoeGermuska]

@iandees when you have time, could you look at this and give it your ultimate OK? I feel pretty good about it (and it's essentially already implemented outside of the dataload process as part of fixing censusreporter/censusreporter#34) but I'd like to make sure you know what's going on.

If you're good, go ahead and merge, etc.

[JoeGermuska]

@iandees don't forget to look at this before we update TIGER, which I'm guessing will be part of the 5-year data load in December?

[JoeGermuska]

I manually ran the updated version of the file in 0ee1439 and native geographies seem to work correctly, so as long as this gets normalized into the data update process in the future, we are good.