Adding column schema attributes

[carolinabtt]

resolves # #1265
docs dbt-labs/docs.getdbt.com/#NA

Problem

Before this PR is merged, we could not identify if the target table have different description and default_value_expression than the model transformation. If the model use the Bigquery API copy_table, and we don't have description and default_value_expression defined in the schema.yml, the metadata will be deleted.

By extending this function _get_dbt_columns_from_bq_table to extract and store the description and default_value_expression from BigQuery's SchemaField objects, we could seamlessly compare the model transformation data with the table destination and decide if replace it or not with the information in the project yaml file.

Solution

One option to avoid deleting the metadata in the target table is using the feature persist_docs = true, but this means an extra step in the process, have to fill all the column descriptions in yml files, and is needed a table.update permission.

With this PR, we avoid this extra step and the metadata is updated with the transformation itself, with the API copy_table or MERGE statement, faster and cheaper.