Add linearization to Mobile Phase Modulator Langmuir Binding

[ronald-jaepel]

This PR stabilizes the Mobile Phase Modulator Langmuir Binding for $c_{p,0} <= 0$.

The isotherm equation:

$$$ \begin{aligned} \frac{\mathrm{d} q_i}{\mathrm{d} t} = k_{a,i} e^{\gamma_i c_{p,0}} c_{p,i} q_{\text{max},i} \left( 1 - \sum_{j=1}^{N_{\text{comp}} - 1} \frac{q_j}{q_{\text{max},j}} \right) - k_{d,i} c_{p,0}^{\beta_i} q_i && i = 1, \dots, N_{\text{comp}} - 1. \end{aligned} $$$

can be undefined for $c_{p,0} <= 0$ for two reasons:

1. The flux implementation is not defined for $c_{p,0} < 0$ due to $c_{p,0}^{\beta_i}$ if $\beta_i$ is a fraction with an even denominator.
2. For $c_{p,0} == 0$ and $\beta_i < 1.0$ the first derivative is also not defined.

Therefore a Taylor series linearization with respect to $c_{p,0}$ has been introduced around $c_{p,0} <= \epsilon$ where $\epsilon$ is a small number.

The CADET-core tests failed with the old formulation for $c_{p,0} <= 0$ and now pass. A full simulation where $c_{p,0}$ dips below zero due to oscillations has been performed to confirm that the linarization works.

Once this is merged we can merge fau-advanced-separations/CADET-Process#163.

[jbreue16]

Looks good to me, but I think we should add a separate test case for the linearization