If we are not conflating spam (content) with spam (quantity), yes. Data hidden via stagnography is not a future IBD issue.
I am not a reactionary here, I just don't like the intentional mischaracterization of a reasonable approach. Creating more of a validation burden on future nodes is the threat, in my estimation.
So, OP_RETURN is fine, as is annex data which don't increase validation burden? And cutting off OP_RETURN and driving those uses to fake pubkeys, which does increase future validation costs, is a threat?