add support to maximize and minimize objective-function

[SimonBlanke]

No description provided.

[SimonBlanke]

This new parameter would determine if the optimum the algorithm is searching for is the minimum or maximum of the objective-function. The API for this could look as follows:

import numpy as np
from gradient_free_optimizers import RandomSearchOptimizer


def parabola_function(para):
    loss = para["x"] * para["x"]
    return -loss


search_space = {"x": np.arange(-10, 10, 0.1)}

opt = RandomSearchOptimizer(search_space, optimum="minimum")
opt.search(parabola_function, n_iter=100000)

opt = RandomSearchOptimizer(search_space, optimum="maximum")
opt.search(parabola_function, n_iter=100000)