bc2pg - simplify

Data loads to existing postgres tables can be done via `bcdata dump | ogr2ogr`, a specific postgres loader isn't really required.

`bc2pg` could be stripped back to just table creation from the catalogue provided metadata:
- read schema from bcdc catalogue
- download a sample of the data
- create empty table based on combination of bcdc schema and data download
- add comments to columns/table
- do not attempt to track the primary keys

Maybe support providing a mapping of {"column_name": "column_type"} to override the detected types - but this is not a priority. Just guess the types / add comments and let the user alter the postgres table definition if any customization required.

Low priority as existing bc2pg works fine apart from minor noted issues - but for v1.0, simplifying makes sense.