Add (optional) actual margin quantile column to completed match.

[max-niederman]

Personally, I've never found the win prediction percent to be very useful post-match. What would be much more useful for me to understand how right or wrong Statbotics was about a match is how extreme the score margin was. Specifically, something like p(margin sampled from EPA model < actual margin | pre-match alliance EPAs). This is of course harder to understand, so maybe it would be best as an optional setting for those who understand the math.

Another thing that could be useful and probably easier to understand would be the score margin error (Actual Margin - Expected Margin).