[native_datafusion] PARQUET_FIELD_ID_READ_ENABLED is not respected

[andygrove]

Describe the bug

There are a number of Spark SQL tests failing related to field ids when reading Parquet files when we use native_datafusion / native_iceberg_compat because these new scans do not fully support field ids yet.

Steps to reproduce

No response

Expected behavior

No response

Additional context

No response