I think putting a hard cap on OP_RETURNs would be the single best non-risky thing to do. However, I think bitcoin adoption will cause blocks to be mostly full a few years from now - and the fees to be much higher. So it would only buy some time & avoid harmful content onchain.

That being said the cost of storage keeps getting lower and lower, but bitcoin block size is static, so all things consider it should only get easier to run nodes from now on.

I agree that financial transaction will likely price out most of the spam in the long term, and block size are static so no storage issue, but the problem isn't there, the problem is in the UTXO set, which inscription such as ordinal creates multiple non used UTXO which nodes need to keep in RAM (so RAM cost are going significantly higher, so node cost more overtime. So I believe one of the most important things is to remove at the consensus level exploit that permit inscriptions to save UTXO set (which is what BIP-110 is mostly doing) now is BIP-110 the good approach, maybe not but at least its working in the good direction.