Indeed it makes little sense that the error rate would be independent of genome length. Even in the case of radiation it's going to depend on the cross section at minimum, which is related to volume, which is related to genome length.
But could there be more error correction since the total storage area is larger?
Does DNA have Hamming codes?
I only have 101 biology, but I think, we don't really know for sure, but at least some repairability must exist based on the fact that women are resilient against some hereditary problems.