Implement AIC and BIC

[adam-m-mcelhinney]

I propose that the AIC and BIC functions are implemented. This would be easy to do on top of the logloss that already exists.
https://en.wikipedia.org/wiki/Akaike_information_criterion
https://en.wikipedia.org/wiki/Bayesian_information_criterion

Note that other implements of AIC are not guaranteed to give the actual AIC value, only an approximation that is missing a constant. I am not sure if that is because of the loglikelihood function or bc of the AIC function.

[mfrasco]

Yes, this is a great metric. We should specify an argument that indicates the distribution type. Also, we should try to make it so that our implementation matches stats::AIC. So we might need to figure out what that constant is.

[timbook]

Is this still open/desired? I'd like to tackle this. Is there also any interest in Mallow's Cp?