Improve regression model performance of the Mapping Saturation indicator

For all OQT regions a total number of 192 models are expected. But only 112 models could be created. With additional checks afterwards the total number of fitted models is reduced to 87 out of 192.
A `RRuntimeError` with an error message will be raised if a model could not be created. Below is a collection of those error message for each model.

- Self-Starting Nls Asymptotic Regression Model
- step factor 0.000488281 reduced below 'minFactor' of 0.000976562
- singular gradient
- Self-Starting Nls Michaelis-Menten Model
- NA/NaN/Inf in 'y'
- step factor 0.000488281 reduced below 'minFactor' of 0.000976562
- Two-Steps-Sigmoidal Model (Tangens Hyperbolicus)
- singular gradient matrix at initial parameter estimates
- Self-Starting Nls Four-Parameter Logistic Model
- step factor 0.000488281 reduced below 'minFactor' of 0.000976562
- number of iterations exceeded maximum of 50
- singular gradient
- NA/NaN/Inf in foreign function call (arg 1)
- Self-Starting Nls Logistic Model due to RRuntimeError
- step factor 0.000488281 reduced below 'minFactor' of 0.000976562
- NA/NaN/Inf in foreign function call (arg 1)