Embracing Discrete Search: A Reasonable Approach to Causal Structure Learning