If you are indeed the coordinator then you have got an edge, you could simply reject all other users in the round I am joining and the attack will work, but again, doing that would be obvious because other users who have valid inputs will be rejected for no good reason and that would indicate the coordinator is attempting the attack.
And what's a "valid" input? Is there such a terminology in their repository? The coordinator can start rejecting certain inputs as "naughty" (which
is part of their terminology btw), and the users are required to accept this with no questioning. Their blacklisting does not indicate sybil attack attempt, as far as they've put it.
Quoting myself from the past:
You register 10 (non-private) inputs, and 1 of them gets rejected, what is your conclusion? To me, absolutely none. Coinfirm might have deemed this one input as inappropriate, or it might be trying to get rid of some coinjoin inputs, so they can use theirs instead and de-anonymize the remaining registered inputs. Who knows. For instance, a 150-input long coinjoin can have its 75 inputs rejected, and replaced with 75 Coinfirm inputs. That leaves the firm with 50% less output set to account for.
Sybil attacks on the p2p network are different, since other nodes won't care if your node is rejecting them or has gone offline, since there is no central coordinator the whole thing is different.
I agree that it is more effective and less costly to execute in coinjoin. The victim of a sybil attack in Bitcoin Core is the client which connects with malicious nodes exclusively, which possess significant computational power. The victim of a sybil attack in Wasabi coinjoin is to connect with just one malicious entity.