N(ext) Runner: improve the coverage of the interface tests

[clebergnu]

We currently have a functional test for N(ext) Runner script implementations at selftests/functional/test_nrunner_interface.py. These are very basic, and can be augmented in areas such as:

The output produced by the scripts should be better parsed, that is:

• [ ] Valid JSON
• [ ] Specific data structure
• [ ] Specific content

Check more interfaces (if available):

• [x] capabilities (should be mandatory)
• [x] runnable-run
• [x] runnable-run-recipe
• [x] task-run
• [x] task-run-recipe

Apply variants for all the current implementations

• [x] avocado-runner
• [x] avocado-runner-noop
• [x] avocado-runner-exec-test
• [x] avocado-runner-python-unittest
• [x] avocado-runner-avocado-instrumented
• [x] avocado-runner-tap
• [x] avocado-runner-golang
• [x] avocado-runner-robot
• [x] avocado-runner-dry-run