Add docs a for exposure generic tests

[erikzaadi]

Depends on elementary-data/dbt-data-reliability#530

Added docs for the new exposure validation tests:

[linear]

ELE-1703 Create generic test for exposure stability

Definition of done: Create a generic test defined per model. The test iterates on the defined exposures and looks for exposures that depend on the model, and checks the columns described in the exposure match the columns in the model (name and data type).

[github-actions]

👋 @erikzaadi
Thank you for raising your pull request.
Please make sure to add tests and document all user-facing changes.
You can do this by editing the docs files in this pull request.

[Maayan-s]

@erikzaadi
This is a technical description of the test.
I think it's better to describe here the use case it solves.
"These tests will fail if there are column changes that break downstream exposures, such as BI dashboards."

Also, it's a good place to say that Elementary Cloud automatically generates exposures, maps the lineage, and adds the test to the models that the exposures depend on.

[ellakz]

do we have this recommendation in all the test types?

[erikzaadi]

saw it in https://github.com/elementary-data/elementary/blob/ele-1703-create-generic-test-for-exposure-stability/docs/guides/add-elementary-tests.mdx#L234

[ellakz]

I think it's important to point out that the test needs to be set up for every model used in the exposure.
Also you might wanna explain a little on how the test works, mainly to communicate why no other setup is needed except this test :)

[ellakz]

Not sure it's possible, but it will be nice if we could do something like:

referenced_columns:
  ref('orders'):
    - name: "order_id"
      data_type: "numeric"
    - name: "customer_id"
      data_type: "numeric"
  ref('customers'):
    - ... 

[erikzaadi]

not sure, and it'll make the "simple" workflow (AKA no need to set the source property since there's only one exposure dependency) obsolute, which IMHO is a shame, it's prolly our 80 vs 20 scenario

[ellakz]

yep I agree. but maybe we can rename "name" and "source" to "column_name" and "model_name"?

[erikzaadi]

Pushed, only difference is node instead of model_name as per our conversation and @haritamar 's comment on the main PR