Contract failed rows support

[tombaeyens]

TODO:
How does this fit into the broader strategy?
What problem or job are we going after?

HF request.

Work out a contract failed rows proposal. Consider the following approach

A failed rows query can be specified. This failed rows query takes the data source name, database name, schema name from the contract as default and allows users to overwrite these properties.

The failed rows query is executed first and results in a table being created. Similar to (and compatible with) dbt failed rows tests.

After the table is created, the contract verification will only measure the number of failed rows (row count of the failed rows table) as a metric.

Ideally the syntax is very similar for dbt failed rows tests as for soda failed rows checks. Eg

columns:
  - name: stage
    checks: 
      - type: failed_rows
        failed_rows_data_source: (optionally override the contract data source)
        failed_rows_database: (optionally override the contract database)
        failed_rows_schema: (optionally override the contract schema)
        failed_rows_dataset: (required specify the name of the failed rows table)
        failed_rows_query: SELECT ... (optionally specify the name of the failed rows table.  if not specified, soda assumes the table already exists like eg when dbt test already created it)

Internal slite spec link https://sodadata.slite.com/app/docs/0vy9umyD92pSML

[tools-soda]

SAS-3260