ERROR:  Invalid data type timestamp without time zone for time bucket
CONTEXT:  PL/pgSQL function _timescaledb_internal.cagg_migrate_execute_refresh_new_cagg(_timescaledb_catalog.continuous_agg,_timescaledb_catalog.continuous_agg_migrate_plan_step) line 17 at RAISE
SQL statement "CALL _timescaledb_internal.cagg_migrate_execute_refresh_new_cagg($1, $2)"
PL/pgSQL function _timescaledb_internal.cagg_migrate_execute_plan(_timescaledb_catalog.continuous_agg) line 31 at EXECUTE
SQL statement "CALL _timescaledb_internal.cagg_migrate_execute_plan(_cagg_data)"
PL/pgSQL function cagg_migrate(regclass,boolean,boolean) line 25 at CALL

### How can we reproduce the bug?

```bash
--DROP TABLE test_table;
create table test_table (
    device_id bigint not null,
    reading numeric not null,
    time_ts timestamp without time zone not null
);

create index test_table_time_ts_idx on test_table(time_ts desc);

--> single node hypertable
select create_hypertable('test_table', 'time_ts', chunk_time_interval => INTERVAL '1 day');

--> (0) data

--delete from test_table;
INSERT INTO test_table (device_id, reading, time_ts)      
    SELECT
                cast(random() * 50 as int) + 1,
                generate_series(1,8467201),
                generate_series(
                    now() - INTERVAL '98 days',
                now(),
                    INTERVAL '1 second'
                );

--> (1) Create a cagg the old way - 'timescaledb.finalized = false'

--drop materialized view test_table_cagg;
CREATE MATERIALIZED VIEW test_table_cagg
WITH (timescaledb.continuous, timescaledb.finalized = false) AS
select
    time_bucket('1 hour', time_ts) as bucket,
    device_id,
    sum(reading) as sum_total,
    avg(reading) as avg_reading,
    max(reading) as max_reading
from test_table
group by device_id, bucket;

--> (2) initiate migration here

call cagg_migrate('test_table_cagg', override => true);