because they have presumed to remove a user-configurable option that impacts spammy transactions on the chain, not just removed it but set such a large default you can now easily jam a small low resolution JPG image into it, and big enough to jam malware binaries into the chain and have them replicate across the network, causing virus checkers to flag errors and interfere with the function of the node. IMO this indicates that there should be a rethink around the subject of mempool filter logic to push it out to user-definable custom filter programs so they are not going to affect users' ability to control what their node relays. it is not clear what is the motivation for this change in mempool policy by core developers but they should just get their hands off the USERS prerogative to filter mempool relaying by making it possible to delegate all filtering to an external executable that accepts new mempool entries and regulates when new blocks will be propagated by the node (so users can set them to not relay a new best block with spam in it, in case within a short time another block that doesn't appears). also, if you thought that client centralization isn't an issue in bitcoin until suddenly now, you obviously haven't read very much of the history of bitcoin's inception or the changes along the way. this has been an issue in some people's minds for a long time. the most recent last time it popped up was because of lightning network being predominantly LND with BTCD running the node client and an improper check that was not in spec that caused nodes to go offline. this was the beginning of the hacks on the datacarrier setting that emerged due to the way taproot was specified, and there was spectacular traffic jams in mempool traffic for months afterwards.

What are you taking about?? There's no consensus change in Core v30. There has always been several ways to store data in Bitcoin blockchain. Bitcoin is censorship resistant, the point it's you can not avoid my consensus-valid transaction to get mined.