Reimplement data initialization process; keeps ending early because of network issues

[MattTriano]

A full installation of TIGER data for all US states involves downloading 10s of GB of data, and with the current implementation, a single network hiccup or other issue will cause execution of the /docker-entrypoint-initdb.d/11_load_data.sh script to end early, and as that script is only called on the initial startup of the postgis_geocoder docker-compose app (with a given docker volume instance) I've had to just tear down the volume, point my VPN to a different server, and try again. Not only is this unacceptable in terms of ease-of-use, but it's also very rude of me to waste Census Bureau bandwidth redownloading the same files.

So my plan is to rework the initialization to download data to a volume/mount_point on the host machine before moving on to the 11_load_data.sh script.

Maybe I'll try to work this into the /Dockerfiles/Dockerfile_postgis implementation, or maybe I'll make it an external shell script that the user has to run. The former seems more convenient, but I will have to learn more new things to go that route.