Support id column mapping

[nicklan]

Currently we only support "name" mapping, but we should also support id

[zachschuermann]

some more context:

• Not really used (but tables converted from iceberg are id mapped)
• Ids need to make it to the parquet reader (duckDB parquet reader supports it)
• Can just fix this in the reader
• Used to be places that checked for id mapping mode and would blow up - col mapping/data skipping support from ryan meant that the refactor removed most of these