Combine linear, ridge, lasso, and elastic net regressor

[lars-reimann]

Is your feature request related to a problem?

We have four different regressors, that are essentially all linear regression with various regularization.

Desired solution

Combine them into one model. This would also get rid of some warnings, since we can internally initialize the correct sklearn model depending on the hyperparameters (e.g. if alpha is zero, just use a linear regression model).

Also take #750 into account. It should closely match the logistic regression classifier.

Possible alternatives (optional)

No response

Screenshots (optional)

No response

Additional Context (optional)

No response

[sibre28]

Done in #843