fix: use column names for database mappers

When writing results where the metrics include 'confusion_matrix', only
the first column name is written. In the case of the confusion_matrix it
is "true_positive". The desired behaviour is to write all column values.

nannyml/io/db/mappers.py

@@ -234,7 +234,9 @@ def _parse(
 
         res: List[DbMetric] = []
 
-        for metric in [metric.column_name for metric in result.metrics]:
+        column_names = [column_name for metric in result.metrics for column_name in metric.column_names]
+
+        for metric in column_names:
             res += (
                 result.filter(partition='analysis')
                 .to_df()[