Fix backwards compatibility of CSTR

[AntoniaBerger]

Fixes #308

Before _constSolidVolume was always zero because the state of init_liquid_volume was declared after the parameters.

To fix this, there is now a variable init_liquid_volume to calculate _constSolidVolume even if the state is not set.
Additionally the calculation of _constSolidVolume has been fixed:

Before we calculated

$$ V^{s} = V^{l}*(1-\varepsilon) $$

which is wrong because we distinguish between the total volume of the reaction $V^{total}$ and the volumes of the liquid and solid phases.
Instead we now calculate

$$ V^{s} = V^{total}(1-\varepsilon) = \frac{V^{l}}{(1-\varepsilon)} - V^{l}. $$

For the last equation we used that $(1-\varepsilon) = \frac{V^{l}}{V^{l} + V^{s}} \Leftrightarrow \frac{V^{l}}{(1-\varepsilon)} - V^{l}$

[jbreue16]

Looks good to me, thanks for catching and fixing this !

[jbreue16]

It should have been
$V_s = V_\text{total} (1-\varepsilon) = V_l \frac{1-\varepsilon}{\varepsilon}$.
Also, we should add a check for zero division.
The corrections are added to PR #339