Establish a data layer

[oleksmarkh]

• move content of database/ (crisimap-data), deprecating that repo (as it doesn't fit into the architecture - now back-end is responsible for keeping the DB notion)
• setup ODBC layer (e.g. Psycopg2 + Flask-SQLAlchemy, see example)
  • create a sample model representing some actual data relation
  • if Very slim, Flask-based RESTful API surface (acting as fixtures/mocks provider) #2 comes first, pass a model through controller to the API surface
  • update/detail the architecture scheme accordingly
• consider introducing migrations (to keep updates streamlined, e.g. https://github.com/miguelgrinberg/Flask-Migrate)