postgres ExcludeConstraint not generating correctly

[notzippy]

Describe the bug Using the postgres dialect the ExcludeConstraint drops the ops when creating the migration file

Expected behavior The constraint should be transferred to the migration file

To Reproduce Using the following model

class Network(Base):
    __tablename__ = 'network'
    network_id: uuid.UUID = sqlalchemy.Column(pg.UUID(as_uuid=True), primary_key=True, default=uuid.uuid4,
                                              server_default=sql.func.gen_random_uuid(), nullable=False)
    networka: str = sqlalchemy.Column("networka", pg.INET)
    __table_args__ = (
        pg.ExcludeConstraint(('networka', '&&'), using="gist", ops={'networka': 'inet_ops'}),
    )

run the command alembic revision --autogenerate the output is generated for that model

    op.create_table('network',
    sa.Column('network_id', postgresql.UUID(as_uuid=True), server_default=sa.text('gen_random_uuid()'), nullable=False),
    sa.Column('networka', postgresql.INET(), nullable=True),
    postgresql.ExcludeConstraint((sa.column('networka'), '&&'), using='gist'),
    sa.PrimaryKeyConstraint('network_id'),
    )

The postgresql.ExcludeConstraint part has dropped the ops, it should have rendered

    postgresql.ExcludeConstraint((sa.column('networka'), '&&'), using='gist', ops={'networka': 'inet_ops'}),

Error No errors occur after generation, but Postgres will fail to create a table when you attempt to upgrade.

sqlalchemy.exc.ProgrammingError: (psycopg2.errors.UndefinedObject) data type inet has no default operator class for access method "gist"                                   
HINT:  You must specify an operator class for the index or define a default operator class for the data type.     

Versions.

• OS: Linux
• Python: 3.9
• Alembic: 1.7.4
• SQLAlchemy: 1.4.20
• Database: Postgres 13

Additional context Not a show-stopper for me, I just needed to manually add in the code.

Have a nice day!

[notzippy]

For reference this is the PR of the addition https://github.com/sqlalchemy/sqlalchemy/pull/5700

[zzzeek]

we dont have reflection for this yet but a straight autogen from the model should be able to work