When we import the data instead of using column store let's do the same as we do for OpenStreetMap and import using hstore to store arbitrary tags (keys, values).

This would then be paired with moving more property normalization logic out of the JINJA and into the layer YAML.

For example:

 jsonb_build_object(
    'name', name,
    'population', pop_max::text,
    'featurecla', featurecla,
    'scalerank', scalerank,
    'min_zoom', mz_places_min_zoom
  ) AS __places_properties__,

Most if not all of that can then be migrated to the YAML.