Support for multiple sem types for nonlinear stimulated data

[Marcus Kalander]

The IIDSimulator already supports multiple sem types for the linear case for a single simulated dataset. This could be added to the nonlinear case as well.

Refer to

trustworthyAI/gcastle/castle/datasets/simulator.py

Lines 273 to 275 in ff135c7

     def _simulate_linear_sem(W, n, sem_type, noise_scale): 
         """ 
         Simulate samples from linear SEM with specified type(s) of noise. 

and

trustworthyAI/gcastle/castle/datasets/simulator.py

Lines 381 to 383 in ff135c7

     def _simulate_nonlinear_sem(W, n, sem_type, noise_scale): 
         """ 
         Simulate samples from nonlinear SEM. 

A difference would be the quadratic case which would be needed to handle somehow.