Final comments on this, to try to clear the air a little -
1. While I agree that the IEEE standards are an excellent reference, they are intended to summarise elements of engineering practice, not to serve as design standards to cover all cases for all time. An example of the changes that have been made in accepted practice since the referenced date of 1986 is the introduction into general service of microprocessor-based transformer protection relays - these allow connection of the CTs on both sides of the transformer to be made in wye, with the required phase shifting and zero sequence current filtering being done internal to the relay.
2. Even prior to 1986, it was standard practice in the world outside of the IEEE standards for CTs to be shared between the transformer differential protection and the REF protection. This was done by the simple expedient of installing a set of carefully selected interposing CTs in the circuit. The transformer CTs on both the Delta and Wye windings were connected in grounded wye. The CT circuit from the Wye winding was then connected to the wye-connected primary of the interposing CTs, with the REF scheme connected to the neutral of this circuit. The interposing CT secondary was connected in delta, providing the necessary phase shift and zero sequence current filter for connection to the differential relay.
3. My experience has been that large generators are always unit connected and consequently are earthed using a distribution transformer/resistor combination. With this situation, earth fault currents are restricted to very low values and provision of a current-based earth fault detection scheme is inappropriate - earth faults are usually detected by neutral voltage measurement.
Anyway, 140668, I hope that all of this has helped to answer your original question, even if has been rather detailed.