[Bug 54413] ws2_32:sock - DuplicateHandle(socket) sometimes look like a socket in test_WSAGetOverlappedResult() on Windows

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=54413

--- Comment #5 from Zeb Figura  ---
(In reply to Zeb Figura from comment #4)
> Safer is probably just to pass WSAGetOverlappedResult() a garbage handle
> value, one that would in practice never be allocated.

Actually we already do that. This test exists because it does, I think, have
value—we want to show that it's not enough for the handle to be valid, it has
to be a handle that ws2_32 knows about. I don't think there's another way to
prove that, we need to use DuplicateHandle().

So instead I sent a patch that reorders the two tests, and puts the more
"destructive" iocp_async_read_closesocket() at the end.

Hopefully that's enough to make this failure go away...