Not well-posed models not always rejected

[Willi Braun]

The following model should rejected as not well-posed:

model B
  Real a;
  Real b=2;
equation
  a = 2*b;
  b = time+2;
end B;

This should be done as soon as possible, but at least when the BackendDAE has been created, where we are waiting until the index reduction and then on the way it get fixed with the aid of removeSimpleEquations.

[Lennart Ochel]

I just tried to check if the number of equations match the number of variables before creating the backend system: ​OpenModelica/OMCompiler#1091. But it seems that there are a couple of models where this isn't the case but we have to accept them anyway.

[Lennart Ochel]

checkModel returns unexpected numbers in some cases. However, it works just fine to compare the number of variables and equations right after creating the backend system.

[Lennart Ochel]

Fixed with 27e2783/OMCompiler.
Test added with 020ac7f/testsuite.