Compare the best model under the same lambdas

[chokkyvista]

Per title - as the augmented Lagrangian is a minimax problem (i.e. min w.r.t. model parameters, yet max w.r.t. lambdas), it doesn't really make sense to always prefer the lower overall-loss (under different lambdas).

As a compromise, we record the lambdas on the fly and reuse them for comparison, as, at least under the same lambdas, a lower Lagrangian implies a better model.