Only if you know what chunks to ask for, which means a more complex data map. If it were pure erasure code then you can ask for missing chunks (do you wait? which is sync) or have to ask for all of them, which means more bandwidth and message complexity.
There is also a tough call as well, if a hacker sends you a scrambled chunk then rebuilding is poly hard to figure there is a bad chunk (can be resolved using name == hash of chunk)
It does not really, it’s a matter of selecting number of replicants to match the n and p parameters. [edit to be clear it’s higher space efficiency I think you mean and that is not a debate I would have with you 
as you would win, but it’s purely space efficiency not redundancy. At least in a p2p network]
This is what needs to be calculated, space is cheap and getting cheaper. This computation and message complexity of restoring a missing chunk will not scale downwards in the way storage costs do.