Redshift: Fix schemas with additionalProperties object getting duplicate columns

[istreeter]

This issue was raised to us on Discourse.

For example, the Snowplow site_search schema has a field like this:

    "filters": {
      "type": "object",
      "additionalProperties": {
        "type": [
          "string",
          "boolean"
        ]
      }
    },

...it would yield two different columns called "filters" in the ddl:

 "filters"        VARCHAR(1024)  ENCODE ZSTD,                                                                                                                                                                                                                     
 "filters"        VARCHAR(4096)  ENCODE ZSTD, 

This only happens if additionalProperties is an object, not a boolean.

This might be related to #104 but it's not exactly the same, so I'm making it a different issue.

[istreeter]

Duplicate of #55