Move json schema functionality from databridge_etl_tools to geopetl

[floptical]

Apparently a json schema function already exists in geopetl, and it does it in a slightly better way by returning it as a dictionary with constraints.

However it doesn't return precision or scale which would allow us to use more precise data type such as smallint, bigint, smallfloat, bigfloat, which can have much better performance than a generic "numeric" type.

1. Remove this extraction method from databridge_etl_tools and merge with geopetl: https://github.com/CityOfPhiladelphia/databridge-etl-tools/blob/master/databridge_etl_tools/oracle.py#L64-L80

2. Add in functionality to use geopetl schema into carto, specifically using our logic here to decide on data types: https://github.com/CityOfPhiladelphia/databridge-etl-tools/blob/master/databridge_etl_tools/carto_.py#L211-L301

3. Update the carto method here to instead used a "etl.extract_table_schema": https://github.com/CityOfPhiladelphia/geopetl/blob/master/geopetl/oracle_sde.py#L20-L23

4. dbtools uploads the json schema here, change it: https://github.com/CityOfPhiladelphia/databridge-etl-tools/blob/master/databridge_etl_tools/oracle.py#L179