Add baseline_scores to actual endpoint

[skatedplrn]

Now we have baseline_target inside actuals data, it require to keep all predicted values on client side baseline_scores: {actual_at: {'r2': 0.2, 'mda': 0.54}}