Stacked area plot for Icecup, extra denominator

[HighDiceRoller]

In some cases we may want to plot dice that effectively have a denominator greater than the sum of the quantities, i.e. there is some amount of "missing" quantity that was filtered out. I'm not sure yet whether there are use cases for this outside of plotting.

Package-side options:

• A variant of reroll() / filter() that removes outcomes but not their contribution to the denominator.
  • A variant of the Reroll object that does the same.
• A variant of reroll() / filter() that also returns the removed quantity.

The trouble is there's not really been use for this so far, and it would touch a large portion of the codebase.

Icecup-side options:

• A separate denominator or filter argument for output() and area().
• Expect outcomes to area to be (actual_outcome, filter).
• Global function similar to matplotlib xlim to set the range.

[HighDiceRoller]

I went with the limit approach for Icecup, since I think it will be rare to want to exclude outcomes in the middle, and at the same time choosing a side to exclude from is beneficial for plotting the CDF / SF. This approach also dodges the normalization question.

There's still an issue with e.g. this situation. Using chart.js stack probably won't work well with CDF / SF.