Fix AuthenticationExtensionsAuthenticatorInputs/Outputs CDDL

[tidoust]

According to the CDDL grammar, after a control operator (called ctlop in the grammar), there can only be a type2 production. In a type2 production, wrapping parentheses can only be used to wrap a type production. tstr => any is a group production. As such, unless I missed something, it cannot be wrapped in parentheses to produce a type2 production. It rather needs to be wrapped in curly braces or brackets.

In other words, from a CDDL grammar perspective, this is an invalid type:

foo .within ( tstr => any )

This is a valid type:

foo .within { tstr => any }

This update fixes the CDDL type definitions that used the .within operator with an invalid type2.

I don't believe the change introduces any semantic change (so more an editorial update than anything else) but note I'm no CDDL expert, I just happened to have been playing with the CDDL grammar recently... If the semantics change, well, the problem still stands: the current definition is not valid CDDL.

---

Preview | Diff

[emlun]

Well spotted, thanks! Unfortunately this still isn't valid CDDL, though. As I recently had clarified on the CDDL mail list, control operators only apply to types, not groups. This PR fixes the right-hand side, but the left-hand side is still a group, so the grammar is still invalid. I will open a meta-PR to complete the change.