Geometry and geography defaults make schema parsing fail with an unactionable NotImplementedError #3975

A geometry or geography column that declares initial-default or write-default cannot have its schema parsed:

Schema.model_validate({"type": "struct", "schema-id": 0, "fields": [
    {"id": 1, "name": "geom", "required": False, "type": "geometry",
     "initial-default": "POINT (1 2)"}]})
# NotImplementedError: Geometry JSON deserialization requires WKT to WKB conversion, ...

Serializing a field that already holds WKB defaults fails symmetrically, as a PydanticSerializationError wrapping the same error.

The cause is that pyiceberg/conversions.py leaves to_json and from_json unimplemented for GeometryType and GeographyType, and those are reached from metadata handling rather than only from value conversion: from_json from _deserialize_default_value (pyiceberg/types.py:377), to_json from NestedField.serialize_model (pyiceberg/types.py:458). (from_json does pass bytes through unchanged; it is the WKT string form the spec mandates in JSON that fails.)

mkdocs/docs/geospatial.md already records WKB/WKT conversion as a deliberate omission, so I am not asking to revisit that. What looks unintended is how the omission surfaces: a NotImplementedError raised from inside Schema.model_validate, naming neither the field nor the default as the reason. The table simply cannot be loaded, and the message gives the caller nothing to act on. An explicit error about geospatial defaults being unsupported would at least make it diagnosable.

Happy to send that patch if it is wanted.

Related: #1818 lists Geometry/Geography as an open item but links #1820, which was closed once #2859 landed the types and the binary and Parquet paths.