The SeedsLoader class abstracts the generic seeding code from the file
dependent implementation. This simplifies how users can add datasets
from new regions, Given that such datasets conform to the
HealthDashboard data requirements (better documentation for the
requirements will come in later commits).

The dataset is specified by the "HEALTHDASHBOARD_DATA_TYPE"
environment variable when seeding as the codebase should also be aware
of the selected dataset.

For now this only adds support for the "sms-sp" dataset from Sao Paulo
city with sensitive data, later commits will include data from Rio de
Janeiro state using only open datasets.