Hi Eyal,
No, sorry but I really don’t think it does. And the reason is that error correction is simply what the protocol is designed to do and the cost of any packet resends is accounted for in the dimensioning of all the resources. In a gigabit copper network (1000BASE-T) the cables (Cat 5e/6) are certified to deliver up to a billion bits per second correctly up to their rated maximum length, i.e. the cost of the error correction is included. Any draw on the resources of the ethernet chip/circuit is already accounted for. So it’s “business as usual” rather than an “un-budgeted overhead”.
As I have pointed out a number of times, the option of an optical fiber physical layer in ethernet networks was principally conceived to increase the distances over which these certified bandwidths could be delivered, and not to address any deficiencies in copper connections used within their certified performance envelopes. From a domestic audio perspective they represent a complex solution to a non-problem.
You’ll find a lot more discussion in this thread which discussed some of the trade-offs of shielded vs unshielded copper cables, and you may also find the dCS Guide to Streaming has some useful background, in particular the discussion about optical networks in section 2.1 “Synchronous vs Asynchronous Interfaces”.