You're more than right. In fact, the paper explicitly mentions 4 correct 5 detect (but fail) doesn't solve the issue because each victim row on either side of

You're more than right. In fact, the paper explicitly mentions 4 correct 5 detect (but fail) doesn't solve the issue because each victim row on either side of the attacker row attains varying levels of multibit (5+) errors. It doesn't fix systems deployed right now, and could be used for attacking hypervisors and other multitenant systems. Might make an interesting class of local privilege escalation attacks to try probabilistically of otherwise correctly-secured systems too.