This change adds more comprehensive benchmarks with a wider-variety of input sizes for g1 and g2 multi exponentiation.