Given the nature of evolution and the fact that perfect Bayesianism is computationally intractable, the thing to be explaining is not how bad our reasoning is, but how good it is.

I'd say there is time to do both.

When looking at biases I find it useful to distinguish between the ones that are the outcome of our reasoning 'just not being good enough' and those where sound reasoning appears to be actively sabotaged. Those cases where evolution spent extra optimization power introducing mechanisms that hinder the formation of correct beliefs due other selection pressures.

Of course we can also spend time explaining how good human reasoning is, either for the sake of saying "Rah! Humans" or for the purpose of developing our own optimisation processes. The two quests for knowledge are hardly mutually exclusive.