TestCase evaluation by an Interpreter shouldn't require the generation and build of the model containing that TestCase.

[mgronover]

Currently we generate and build the model to evaluate the containing TestCases.
This is done in the way the JUnit-Test-Framwork expects it and is also working for the CI (AFAIK).

For DSL users implementing new TestCases this approach slows them down, because it requires to build the TestCases.

A new Run Configuration, having an Executor that calls the Interpreter to evaluate the TestCase, get the result and update the MPS-JUnit-ToolWindow just like it was done before, would solve this situation.

So currently we are focusing on the IDE usage here.

What is to be expected:

• the run configuration should check and warn if the interpreter was not build
• the interpreter tests should always be executed in the same process
• the configuration panel only suitable options (no more than required!)