[FEA] Ability to control field ID used for columns during Parquet write

[jlowe]

Is your feature request related to a problem? Please describe.
Parquet columns include a field_id identifier, and Spark 3.3 recently added improved support for this feature. See SPARK-38094 for details. In order to support this Spark feature in the RAPIDS Accelerator, we need the ability to specify the ID to use for a column's field_id when the Parquet file is written and these field IDs have been specified in the schema to write.

Describe the solution you'd like
column_in_metadata could have two additional fields and a new set method, e.g.:

bool _has_parquet_field_id = false;
int32_t _parquet_field_id;

column_in_metadata& set_parquet_field_id(int32_t field_id)
{
  _has_parquet_field_id = true;
  _parquet_field_id = field_id;
  return *this;
}

During the Parquet write, if a column_in_metadata indicates it has a field ID setting then the specified field ID is used when encoding the column metadata in the Parquet footer, otherwise a field ID is left unspecified.

Additional context
See the field_id in the Parquet Schema Definition