feat: Extend the SQL connector API to support adapting to primary key changes

[edgarrmondragon]

Feature scope

Targets (data type handling, batching, SQL object generation, tests, etc.)

Description

Right now the flow IIUC goes like this:

1. SQLConnector.prepare_table is called
   1. If the table doesn't exist or load_method is overwrite a new table is created.
   2. Else, SQLConnector.prepare_column is called for each of the table's columns.
      1. If the column doesn't exist, it's created.
      2. Else, an attempt to update the column type is made, if necessary.

So, there's a probably missing SQLConnector.prepare_primary_key API, to be called after columns are updated that updates the table primary keys if necessary.

Related:

• Primary key constraints do not get updated when primary keys change sqlalchemy.exc.IntegrityError(snowflake.connector.errors.IntegrityError) MeltanoLabs/target-snowflake#273