[🚧] new multi tests

[wolfv]

• decide on adding pip_check: True / False to PythonTest.
• Reuse previous environments, ignoring pip
• Decide on serialization format, should we keep run_test.py or serialize to a JSON?
• For the Commands test, should we serialize (potential) if selectors so that people can have optional dependencies based on their test platform? E.g.

requirements:
  build:
     - if: build_platform == linux-64
       then:
         - arm-emulator

[wolfv]

This is how we are currently serializing the 3 different test types:

PythonTest:

• run_test.py (this contains the imports that are listed as import bla..., one per line)

CommandsTest:

• test_time_dependencies.json that contains a dictionary with runandbuild` keys
• run_test.sh / run_test.bat. If it's an arch'd package, either the .sh or .bat. If it is a noarch package we write both files.

DownstreamTest:

• downstream_test.json: { "downstream": "package_name" }.

[wolfv]

The Python tests are now serialized to a python_test.json file instead.