Joining a large table against a tiny table does not push down the extent of the tiny table into the big table

Not sure if this is something that was supposed to work already or not, but the following example doesn't execute for me without downloading all of Overture buildings:

```python
import sedona.db

sd = sedona.db.connect()

sd.read_parquet(
"s3://overturemaps-us-west-2/release/2025-08-20.0/theme=buildings/type=building/",
options={"aws.skip_signature": True, "aws.region": "us-west-2"},
).to_view("buildings", overwrite=True)

sd.sql("""
SELECT label, ST_SetSRID(ST_GeomFromWKT(wkt), 4326) AS poly
FROM (VALUES
('A', 'POLYGON ((-73.22 44.03, -73.22 44.01, -73.17 44.01, -73.18 44.03, -73.22 44.03))'),
('B', 'POLYGON ((-73.14 44.03, -73.14 44.01, -73.08 44.01, -73.09 44.03, -73.14 44.03))'),
('C', 'POLYGON ((-73.19 43.99, -73.18 43.96, -73.11 43.96, -73.12 44.00, -73.19 43.99))')
) AS t(label, wkt)
""").to_view("join_table", overwrite=True)

sd.sql("""
SELECT join_table.label, buildings.id, buildings.geometry
FROM buildings
JOIN join_table
ON ST_Contains(join_table.poly, buildings.geometry)
""").to_memtable()
```

I am not sure if the solution here is in the spatial join or whether the solution is to start implementing Statistics propagation.