Support nested column descriptions in BigQuery

[tanghyd]

I would like to add nested column descriptions to SQLMesh models. If running on BigQuery then the data type would be a RECORD type.

For example I don't just want to document the my_record column, I want to document its nested fields my_record.field_1 and my_record.field_2. All 3 must be described, not just the top level.

This is especially helpful for wide tables that are pre-joined for performance reasons at query time.

I expect the syntax may look like this:

model (
     name my_dataset.my_table,
     description "Contains rows.",
     column_descriptions (
         my_record = "Description of the my_record struct."
         my_record.field_1 = "Description of field 1."
         my_record.field_2 = "Description of field 2."
     )
 );

This functionality is present in dbt but is not available with sqlmesh. See below for how this is documented in a schema.yml file:

      - name: my_record
         description: Description of the my_record struct
         data_type: struct
       - name: my_record.field_1
         description: Description of field 1.
         data_type: string # example
       - name: my_record.field_2
         description: Description of field 2.
         data_type: array<string>  # example

This feature should work regardless if the column is a RECORD (aka STRUCT) or a REPEATED RECORD (aka ARRAY of STRUCTs).

Also see: https://tobiko-data.slack.com/archives/C044BRE5W4S/p1739773563724809?thread_ts=1739773563.724809&cid=C044BRE5W4S