Issue with empty arrays and iterators in the NF

[Francesco Casella]

Please check the ​Modelica.Fluid.Examples.Tanks.EmptyTanks model. One more variable than equations is reported.

Inspection of the flattened models shows:

tank1.ports_mC_flow[1,:] = array(actualStream(tank1.ports[1].C_outflow[$i1]) for $i1 in 1:0) *
      tank1.ports[1].m_flow "Trace substance mass flow";

which shouldn't be there in the first place. It also shows

der(tank1.mC_scaled) = tank1.mbC_flow / {};

which shouldn't obviously be there - BTW, there is (correctly) no declaration of tank1.mC_scaled in the flattened model.

[Per Östlund]

The issue of empty array equations not being removed should be fixed in 5c30da0, but it doesn't affect the issue of the model being unbalanced. Which I guess is expected, since the backend should just ignore those equations when counting.

[Per Östlund]

With the fix in 9f7f5f7d the flat model is now balanced. The model now gets as far as the code generation, which fails because the model contains cardinality which isn't yet handled by the NF.