Contract error messages use internal names of predicates

[97jaz]

See #14 for an example. This probably has to do with the object-name of the procedures in question. Maybe I just shouldn't impose contracts except at the very public-private boundary.

I've run into some really annoying issues with contracts in this project (which I could probably work around if only I understood the syntax system better). For example, the dual role of struct ids as constructor (procedure) and match-expander (syntax) does not play well with recontract-out.

(This issue is more a reminder to me to avoid this problem in the newer datetime library.)