Bug: BigQuery Repeated field should be ARRAY[<data_type>] in model yml

[Thrasi]

Given the model testModel:

version: 2
models:
  - name: testModel
    config:
      materialized: table
      contract:
        enforced: false

SELECT
  [ STRUCT( 1 AS nested_field_1,
    "a" AS nested_field_2),
  STRUCT( 2 AS nested_field_1,
    "b" AS nested_field_2),
  STRUCT( 3 AS nested_field_1,
    "c" AS nested_field_2) ] AS repeated_record,
    [1,2,3] AS repeated_int

dbt-osmosis yaml refactor testModel

produces the configuration:

version: 2
models:
  - name: testModel
    config:
      materialized: table
      contract:
        enforced: false
    columns:
      - name: repeated_record
        description: ''
        data_type: RECORD
      - name: repeated_record.nested_field_1
        description: ''
        data_type: INT64
      - name: repeated_record.nested_field_2
        description: ''
        data_type: STRING
      - name: repeated_int
        description: ''
        data_type: INT64

Setting config.contract.enforced: true and rerunning dbt
gives us a contract mismatch:

Column 'repeated_record' has type STRUCT<nested_field_1 INT64, nested_field_2 STRING> which cannot be coerced from query output type ARRAY<STRUCT<nested_field_1 INT64, nested_field_2 STRING>> at [9:5]

This does not match the BigQuery schema. The repeated_record column should have data_type: ARRAY instead of data_type: RECORD
For bigquery columns we need to take into account the mode of the column which can be REQUIRED, NULLABLE or REPEATED when specifying the data_type in the yml

For basic datatypes like repeated_int, the correct data_type there is data_type: ARRAY<INT64>