Enable different identifiers in YAML-File than hosts

[bytinbit]

@ubaumann raised a good issue in his PR #34 : Currently, the host is the identifier in each test case of a YAML file. This is unpractical and rigid, it would be better to allow other data entries as identifier, such as tags in VRF tests as demonstrated by @ubaumann .

However, this requires a larger refactoring of the code base. PR #34 suggests a solution, but it has several drawbacks, one of which is that the context has to be loaded twice just so that the data fro the YAML can be unpacked, i.e. enhanced with data.

Several solutions have to be assessed before the code is refactored:

• Enable some sort of preprocessing that enhances the YAML test data with inventory data that are required for the tests. Idea: Pre-Load YAML, correlate it with inventory, write new YAML with updated test_data that is finally used for the whole testing process with nuts.
• Introduce a new class that is an interface between the YAML file and the inventory

Please add further ideas in comments.

Related: #27

[bytinbit]

Idea: build CLI, do preprocessing of yaml file, then invoke pytest internally: https://docs.pytest.org/en/latest/how-to/usage.html#calling-pytest-from-python-code

[ubaumann]

In my opinion we would lose a lot of pytest features and nuts would not need to be a plugin anymore

[bytinbit]

it's an idea worth trying