Multiple constraintSpecs with the same `@ident` should be invalid

[martindholmes]

With @sydb / ATOP: We are considering the possibility of an ODD file which contains multiple <constraintSpec> elements with the same @ident, and what should happen when processing such an ODD. We believe that it should be invalid to have the same @ident value for more than one <constraintSpec> in the same ODD file. We cannot think of a use-case where this makes sense or provides any functionality, and we can think of lots of potential problems in trying to process a customization against such an ODD. We would like to add a <constraintSpec> to TEI which invalidates this situation.

[martindholmes]

The current unique-constraintSpec-ident constraint almost handles this, but not quite; the predicate [ @mode eq 'add' or not( @mode ) ] should be deleted.