Fix miscellaneous invalid hyperparameter options in the configuration dictionary. And add n_jobs=1 when possible.

[perib]

I went through some logs to check for classifiers, transformers, and selectors issues. Those seem to not generate invalid hyperparameter combinations. I have not checked regression yet.

It would also be good to check for extreme values that may be unreasonably slow or blow up memory. For example, when PolynomialFeatures was set to a high degree, it would often create extremely large matrixes that would blow up memory and potentially crash the system.

Also need to add n_jobs=1 to parameters that have an n_jobs parameter.

[perib]

More info on parallelization with sklearn https://scikit-learn.org/stable/computing/parallelism.html#parallelism

n_jobs=1 should prevent parallelizable models from using more cores than we want.

[perib]

There are some parameters included that should not be learnable because they are not directly related to model fit.

I don't think 'tol' should not be a trainable parameter rather, it should be fixed to something like 1e-4

copy_X should not be a trainable parameter; remove and leave it default.

Precompute should not be learnable; leave it at default.