Constrain uniqueness of column name and move geocoding column creation

[nllong]

Any background context you want to provide?

In #4146 there were two of the exact same columns (geocoded_county) created during the geocoding column creation. They were created within 0.003 seconds of each other. I attempted to recreate the issue but was unable as this was probably a race condition. To fix the issue, a db admin will need to remove the redundant column manually from the database. @RDmitchell - I did this for your LBNL 405 organization if you want to continue testing.

What's this PR do?

To prevent this from happening in the future 2 things were done:

1. I moved the initial call of the geocoded column creation to before the celery_chain call. Therefore, if this was a parallelization race condition, then the columns should exist when the geocoding runs the get_or_create call.
2. I added a constraint on the Column to enforce a unique name across the column_name, org, table_name, is_extra_data, and units_pint columns. Using the production database, there were no conflicts when running.

I debated on adding the units_pint to the constraint and ended up adding it because I can see the use case where there are two mapping profiles where one set of data are coming in as m2 and another is coming in as ft2; both of which should be allowed to exist. IF we remove units_pint then the production database will have to resolve some naming conflicts which exist due to old orgs not having the units_pint and defaulting to None. @haneslinger -- I know we discussed this constraint in the past but I don't remember the context.

Other items in the PR:

• Update shapely to 2.0.1. It was segfaulting on WKT translation with Python 3.9 on osx
• Remove unneeded print statements

How should this be manually tested?

• verify unit tests
• Add geocoding MapQuest key
• With celery running (non-eager mode)
  • import a file with address (no lat/long) and verify additional geocoding fields (e.g., geocoded_neighborhood)
  • in the inventory page, select a few properties and then select geocode
  • via swagger, go to the geocode_by_ids and add in a property id and org.

What are the relevant tickets?

#4146

Screenshots (if appropriate)