pandas groupby apply numpy.polyfit to tens of thousands of groups quickly

Question

For monthly records of some count n for a given set of groups, I\'d like to fit a basic, order-1 regression line each year\'s results, basically calculating how