Determine how/whether we should explicitly enumerate functions that don't propagate taint

[mlevesquedion]

From @PurelyApplied on #292:

• Would it be better to explicitly enumerate the "no taint propagated" methods and indicate error here if an unemumerate function is found?

• I don't expect there to be much churn in the stdlib without us noticing, but it could be a good defensive measure to make sure we're not skipping over some functions without deliberately having excluded them and/or end up with a function signature that should be matching a summary but isn't due to a bug.

(Link to thread)