Census tracts surrounded by DACs should also be considered DACs

[emma-nechamkin]

This issue seeks to address "donut holes" in our DAC map. This will involve the following:

1. For all census tracts that have at least one DAC neighbor, calculate the share of neighbors that are DACs
2. Over a certain threshold, those tracts will also become DACs

In analysis, I did this analysis first within state, and then, nationally for all tracts with 50% or more DAC neighbors within state.

Some caveats:

1. First, we will have to move up some of the geocoding / reading geojson work from etl_score_geo to etl_score, since there are now parts of the score that depends on neighboring tracts
2. The exact neighbor share used depends on final analysis; we will start with a threshold of 85%

[emma-nechamkin]

Waiting for CEQ approval to implement this