Custom searcher ignoring max_cost_total

[danrgll]

When utilizing a custom searcher, the parameter max_cost_total is currently being ignored.

To make clear what scenario I mean, see this example:

def run_pipeline(**config):
    return {"loss": 1, "cost": 3}

search_space = SearchSpace(**pipeline_space)
optimizer = BayesianOptimizer(search_space)
neps.run(run_pipeline, searcher=optimizer, max_cost_total=100)