Take best of a set of CDFs / SFs

[HighDiceRoller]

Unfortunately this isn't really a "physical" operation, you're basically making the best choice of the individual distributions for every threshold, but it doesn't quite feel like a probability distribution in the conventional sense.

[HighDiceRoller]

Possible naming:

• highest_threshold or threshold_highest
• maximum_sf / minimum_cdf
• highest_by_quantile
• highest_by_probability
• highest_cumulative
• pointwise_highest
• pointwise_max

One recurring issue is the difference between "highest" and "max". I didn't want to name a function max because it would shadow Python's built in max. There's also the precedent in gaming manuals, Roll20, AnyDice, etc. for saying "highest" rather than "greatest" or "max". We do currently have "max_outcome". We could say that "max" is static and "highest" is probabilistic, though this still leaves this pointwise case in kind of a middle ground where it's static in the sense that we're not actually rolling all of the argument dice, but it's probabilistic in that sense that the result is a die.

[HighDiceRoller]

Went with pointwise_max.