OutputInstance

defines these:

apply<U>(func: (t: T) => Promise<U>): Output<U>;
    apply<U>(func: (t: T) => OutputInstance<U>): Output<U>;
    apply<U>(func: (t: T) => U): Output<U>;

Shouldn't they be?

apply<U>(func: (t: T) => Promise<U> | OutputInstance<U> | U): Output<U>;

Otherwise if

func

is something like

string => number | Promise<number>

then it returns

Output<number | Promise<number>>

rather than

Output<number>

. Unless that was intentional?