Process of using and fixing MLVM tests revealed a number of design decisions, which either are bad or do not work well, or seem too much tangled for other people.

This RFE was created to discuss the issues and the way to fix them.

The goal of this RFE is to refactor MLVM into a framework, which has a good design (or at least doesn't have obvious design flaws) and can be understood well and easily used.