the default engine parquet reader should convert the list schema to be named element

[nicklan]

This matches the parquet spec and means that reading parquet via delta will have arrow data that matches the arrow schema we generate from the schema we read out of the delta log.

This should likely be done in the arrow_utils.rs stuff where we re-order/filter everything. This might be complex, especially in the case where this rename is the only thing that needs to be done. We'll need to capture that so that the reorder_struct_array call notices that this needs to happen.

Once fixed, enable the parquet-all-types-legacy-format golden test.