Create a job which consumes the output of bhr_collection.py to output a history of hang counts over time for the top hanging stacks

[squarewave]

This will help us more quickly identify regressions, which should help us actually get them fixed. I expect we'll only want to do this for the top, say, 500 hang signatures, as creating this for the very long tail of hang signatures would be substantially more time consuming and unlikely to be worth it.