Serialize UUIDs or show a better error when panicking on object columns

[jacopofar]

Description

When reading UUIDs from a Postgres database using read_database they get converted to Python UUID objects which causes a panic on some operations. As mentioned in #7175 having a UUID type could solve the problem. In any case, I think having an error reporting which column caused the issue would make troubleshooting a lot easier.

for example:

df = pl.read_database(
        "SELECT '11111111-1111-1111-1111-111111111111'::UUID",
        sqlalchemy.create_engine(
            'postgresql+psycopg://localhost'
        ),
    )
df.write_parquet("error.parquet")

this code will cause an error:

pyo3_runtime.PanicException: called `Option::unwrap()` on a `None` value

when invoking write_parquet because the type of the UUID column is "object". Replacing ::UUID with ::TEXT in the query avoids it.

I wouldn't call this a bug, it's only hard to troubleshoot as the message doesn't make it clear what the problem is (for example seeing the column name would be useful).