Changed the training samples and added BayesInf and additional models for BMC

Finished the example for BayesInf for the user guide and fixed some small bugs for MCMC with a single input.
Started working on the user guide for BMC, but issue in AL with changes in the model

See issue #46, voronoi exploration was removed for v1.1.0, but should be reactivated in future releases. The corresponding tests were changed accordingly.

All combinations of tradeoff-exploration-exploitation run through with SeqDesign. Reverted BayesInf to the current develop branch

example_analytical_function_testSequential runs through all the sequential options and contains notes on which do not work.
Choose_next sample now uses only one set of candidate samples

User guide startet for models, surrogate training, sequential training, postprocessing and BAyesian classes. The guide now contains simplified UMLs (for BayesInf not yet done).

Errors thrown by the tests were resolved where possible.
If they could not be resolved and issue was created for them.
For the sake of testing the pipeline, the unresolved tests were commented out.

Changed scipy, tqdm, statsmodels

An example 'principal_component_analysis' was added to specifically test this functionality. It uses the analytical model.
Fixed two if-calls in `pca_transformation` to match older code and the content of the function.