Towards unbiased benchmarking of evolutionary and hybrid algorithms for real-valued optimisation