That is why we have the minimum fee in place. It makes spam attacks economically unprofitable and drains bitcoin from the spammer.
We have run for 6 years without a fee market and no spam attack has taken place.
There's no minimum fee dictated by the protocol. You can send 0-fee transactions if you wish.
...and in times of congestion, your 0-fee transaction will not get included in any blocks. Fine. But what happens when everyone (e.g. more than 4-7 per second) is including a fee? Is it reasonable that nobody is granted access to the system but those prepared to pay 0.1 BTC? 1 BTC? 10 BTC? The central question is whether or not a protocol limited to 4 tps is reasonable. I reply with an emphatic 'NO'.
As for no spam attack, what else would you call the "stress tests" we had?
Let us amend that to 'no effective spam attack'.
But small blocks only make it cheaper for an attacker to clog the system. Large blocks make it more expensive to clog the system. With large blocks, the attacker must create more transactions, each accompanied by whatever minimum fee the miners on average set for their individual policies for fee threshold for block inclusion.