In the Originator-ID, *both* the public key and the key selector
(which is supposed to be for hiding the public key) are
provided.
This starting premise is incorrect. The key selector may be used in
this way but it is not required.
This is the kind of thing the suggests a great deal of confusion over
what the key selector is for. The draft doesn't give me any help as
an implemetor about what I'm supposed to do with such a
construct.
What you do with it? I assume you mean what is a recipient supposed to
do with this information? Why use it to identify the public key used,
of course. The recipient needs to retain the information so that it can
be included in future correspondence and, in fact, the recipient may use
it as a part of the index used to store the public key in a database.
As an originator, you (or your implementation on your behalf) simply
chose a value, store it with the key pair, and use it in originator-id:
fields to indicate which key you've used.
Jim