Hardcoded tolerance

[Robbybp]

Does k_aug have a tolerance option? For the tolerance used e.g. to check strict complementarity? It seems like it should, instead of hardcoding a tolerance of 1e-8 in the adjust_mu_primal.c file.