Feature request: algorithm breakdown for multiple seasonal arima

If possible, it would be super helpful to see an algorithm breakdown for multiple seasonal arima, similar in style to what you've done for the arima.ipynb in this repo. Happy to help develop with you. R's smooth package might be a good reference to use (python's statsmodels does not handle multiple seasoanlities).