On self discounting, it has more to do with how you define your curve. They seem to be defined as a single list of instruments, like a self discounting curve. Whereas with multiple curves (OIS / Libor 1m / Libor 3m, etc), you would need to provide the instruments that define the inter-dependency of the curves (like 1m-3m basis swaps, etc) and you kind of need to build them all at once, since they are interdependent (swap rates vs 3m are using both the 3m and the OIS curve).
Anyway, I have never looked at Quantlib myself, so all of this may be limitations of the underlying library. But if you present a yield curve product to an interest rate professional, the question will come up.
On index names, I think it should be data more than hard-coded enums. There is nothing in the calculation that should explicitly refer to the name of an index. It should be all about daycount fractions, calendars, and other conventions.
Anyway, I have never looked at Quantlib myself, so all of this may be limitations of the underlying library. But if you present a yield curve product to an interest rate professional, the question will come up.
On index names, I think it should be data more than hard-coded enums. There is nothing in the calculation that should explicitly refer to the name of an index. It should be all about daycount fractions, calendars, and other conventions.