Yes, exactly, which is discussed in the paper. The goal of gain of function research isn’t about “let’s make viruses that do crazy shit,” it’s a method of hypothesis testing. When you can successfully add (or remove) stuff then you can test hypotheses that would otherwise not be testable.
Introducing feature X (regardless of effect) allows you to study X.
You do it to learn more about X, not because you particularly want to engineer something more virulent or lethal.