Make an RFC about algebraic data / complete matches

[jeapostrophe]

In the Racket1 ecosystem, there are multiple incompatible ways that people try to define datatypes with a fixed number of variants and get warnings in consuming code that does not handle all cases. One of the earliest is define-datatype from plai. There should be a standard way to do this pattern that integrates nicely with standard structs/classes.

[jackfirth]

This should probably integrate with match somehow, which (I think) means match will need some way to understand when a set of match clauses is exhaustive.