Cdf for multinomial

Absolutely go with LeBrun if he’s thought about it extensively.

(My derivations only evaluate the incomplete gamma function at positive real values.)