Just curious, why didn't you start with picking a reference recording and then trying to match that the closest possible way with different libraries combined, and do a poll which one we think got closest?
I'm not gonna vote because I don't feel qualified, don't have the required familiarity with Beethoven's music and don't have the attention span to compare that amount of music in a fair way, but the whole methodology of this test is confusing me a little. If you only care "which sampled orchestra fits best out of the box", then... ok... But I'd have been much more interested in "how to get the closest to reference recording X by combining whatever libraries necessary" if I was doing such a test. If you had a ref-track that you are trying to match, I'd imagine you already would have been able to throw out a ton of options on your own and not need to ask people to compare so many versions of the piece to each other in the first place. Maybe narrow it down to the three best or so first on your own. And unless you provide a reference recording, who is to say that the vote won't be strongly influenced by personal preferences for the overall sound of certain libraries?
I'm sure you haven't put that ton of work into this on a whim and I must be missing something (me being a noob and all), but like I said, I'm a little confused by this whole setup...