Implement primary key inference for model nodes

[MichelleArk]

Housekeeping

• I am a maintainer of dbt-core

Short description

Now that we support both constraints and tests on models, dbt can infer a primary key with the following logic:
1. Columns participating in a model-level primary key constraints
2. Columns with primary key constraints
3. Columns with unique and not_null data tests
4. Columns with enabled unique or dbt_utils.unique_combination_of_columns data tests
5. Columns with disabled unique or dbt_utils.unique_combination_of_columns data tests

Let's start by implementing this as an method to serve as a building block for downstream uses cases such as discovery (e.g. through the manifest), and maybe even sampling?

Acceptance criteria

• ModelNode.infer_primary_key exists!

Suggested Tests

• cases in description are covered

Impact to Other Teams

N/A

Will backports be required?

nope

Context

No response