Random Benchmarks for Evaluating Logic Optimizers