Directly scanning metadata file

[samansmink]

Partial solution to #10.

We now allow directly passing a metadata file. We check if the path is a json file, if it is we just scan that directly. This will allow users to access the various iceberg catalogs through an external tool, then pass the path to the latest to duckdb.
Until we have proper support for iceberg catalogs, this will allow actually using the iceberg extension on systems that don't produce a version-hint file:

SELECT * FROM ICEBERG_SCAN('../<metadata_file_name>.json');