DataSealer dependencies

[Cantor, Scott]

> That's an option, but one I deliberately avoided in the initial implementation
> since the service doesn't really start a session until the ticket is validated on
> the back channel. It wouldn't break the protocol to do it on the front channel,
> so definitely an option, but it will have storage consequences in practice. It's
> quite common for tickets to be granted and not validated due to
> configuration problems and environmental blips.

Sure, but I don't think that's substantial volume. It's not any different with SAML or any other protocol, and in the case of a POST, we wouldn't even know whether the session was ever actually started. I don't think it's necessary to design for the edge case, particularly when it's configuration related.

-- Scott