Handle DedupedColumnCollection in db_indexes

This should fix the `scripts/check_model.py` script and the database
indexes github workflow.

Accessing index columns has changed: index.columns gives access to a
DedupedColumnCollection instead of a ColumnCollection.
Columns can be accessed by key; keys are accessed via keys() attr.

Ref:
https://docs.sqlalchemy.org/en/14/changelog/migration_14.html#select-objects-and-derived-from-clauses-allow-for-duplicate-columns-and-column-labels
https://docs.sqlalchemy.org/en/14/core/sqlelement.html?highlight=dedupecolumncollection#sqlalchemy.sql.expression.ColumnCollection

scripts/check_model.py

@@ -19,7 +19,7 @@ IndexTuple = namedtuple('IndexTuple', 'table column_names')
 
 
 def tuple_from_index(index):
-    columns = tuple([getattr(index.columns, c).name for c in dir(index.columns) if not c.startswith('__')])
+    columns = tuple([index.columns[key].name for key in index.columns.keys()])
     if len(columns) == 1:
         columns = columns[0]
     return IndexTuple(index.table.name, columns)