Refactor check valve with spline interpolation

[AntoineGautier]

This addresses #1937.

[AntoineGautier]

@mwetter This is ready. Can you suggest a name for an external review?

[mwetter]

@AntoineGautier : I am not sure we approached this the right way, in part because I don't understand your implementation (see inline comment).

Wouldn't it be simpler if we were to compute $dp$ from $\dot m$? This would make it simpler because both components have the same $\dot m$, so we can simply add their resistances. For the fixed resistance the evaluation is easy (with the flow function). For the actual valve, we can make a linear behavior for $\dot m \le 0$ and quadratic for $\dot m \ge \dot m_0/2$. In between we use a polynominal that fits the first and second derivatives at these points. The coefficients of the polynomial can be computed as parameter expressions.

[AntoineGautier]

@mwetter With 11097a8 the function calls are now included in the if else clauses, min and max are removed and parameters are used where possible.
I've also updated the documentation for the derivatives of the basic flow functions as discussed. Note the important information that these derivatives are formulated under the assumption of a constant flow coefficient. This is consistent with the annotation derivative(zeroDerivative=k) in basicFlowFunction_dp. This also means that if a model features a varying flow coefficient (e.g., any actuator model), these derivative functions will not be used by the tools, which will rely on automatic differentiation instead.

[mwetter]

@AntoineGautier : Thanks for the changes. This addresses my concerns that we discussed on Friday.