Ignore/ raise warning when "extra" parameters are provided to certain functions.

[rickvanveen]

Is your feature request related to a problem? Please describe.
The parameters between activation functions and solvers change depending on the type that is selected. When constructing a grid for sklearn's gridsearch it would be very convenient if "extra" parameters would be ignored instead of raising an error.

Describe the solution you'd like
Raise warnings instead of error when "extra" parameters are provided

Describe alternatives you've considered
Provide custom function that deletes any "extra" parameters from the grid. This would need to be called before passing the grid to gridsearch. Less convenient solution.

Additional context
N/a