Completely wrong results in system with conditional equations

[casella]

Please check Modelica.Electrical.Analog.Examples.DemonstrateLightning.bmo. After the event at time = 1e-5 s, the results of the MTK simulation differ wildly from the reference result.

@AnHeuermann simulated the Base Modelica code with OMC and got results consistent with the reference results, so it seem that the flattening and Base Modelic exporting process is OK.

The issue is likely due to the two conditional equations

    if time < 1e-5 then
      'lightning1.signalSource.y' = 0.0;
    else
      'lightning1.signalSource.y' = 1e5 / 'lightning1.signalSource.eta' * (exp(-(time - 1e-5) / 'lightning1.signalSource.tau1') - exp(-(time - 1e-5) / 'lightning1.signalSource.tau2'));
    end if;

and

    if time < 1e-5 then
      'lightning2.signalSource.y' = 0.0;
    else
      'lightning2.signalSource.y' = 1e5 / 'lightning2.signalSource.eta' * ((time - 1e-5) / 'lightning2.signalSource.tau1') ^ 5.0 / (1.0 + ((time - 1e-5) / 'lightning2.signalSource.tau1') ^ 5.0) * exp(-(time - 1e-5) / 'lightning2.signalSource.tau2');
    end if;

as all other equation are simple linear DAEs.

Possibly related to #99.