Normalize seq_name in pgSerializer getColumnsInfoQuery

[mshick]

Prevents a failure to match sequences when introspecting non-default schemas by normalzing the seq_name column.

The seq_name, assigned as identityName in the Postgres introspection is treated as though it is an unqualified column name in multiple locations in the code.

This leads to several bugs when using a non-default schema, notably, any push to a database after the first push will generate a bad change set that attempts to alter the schema, often by trying to drop the sequence which failed these checks.

Concretely, given a push for my-schema.todo, the seq_name for an id column returned from the introspection query would be "my-schema".todo_id_seq, whereas a default public.todo table would return the unqualified todo_id_seq. All of the code is expecting and handling the default case only, for instance, concatenating the schema name with the seq_name generating keys like my-schema."my-schema".todo_id_seq and testing those against sequences that would have the key my-schema.todo_id_seq.