[core] Fix Iceberg metadata parsing error by excluding null doc field from JSON serialization

[KangJiJi]

Purpose

When using Paimon table with metadata.iceberg.storage option to read metadata as Iceberg table, Iceberg fails to parse the metadata JSON with the following error:

Error while reading table: iceberg, error message: Failed to parse Iceberg table schema 
with the following error: [json.exception.type_error] type must be string, but is null

This happens because Paimon serializes the doc field as "doc": null when the field description is not set. According to Iceberg spec, the doc field is optional. It is preferable to omit it rather than set it to null when not present.

This PR adds @JsonInclude(JsonInclude.Include.NON_NULL) annotation to the doc field in IcebergDataField, so that null values are excluded from JSON output.

Before:

{
  "id": 1,
  "name": "column_name",
  "required": false,
  "type": "string",
  "doc": null
}

After:

{
  "id": 1,
  "name": "column_name",
  "required": false,
  "type": "string"
}

Tests

• IcebergDataFieldTest#testDocFieldSerializationWithNullValue
• IcebergDataFieldTest#testDocFieldSerializationWithNonNullValue

API and Format

No API changes.

Documentation

Iceberg

[JingsongLi]

+1 thanks @KangJiJi