Consider Drell-Yan process

$$

P + P' \to l + l^{\dagger}, \qquad (1)

$$

where $P, P'$ are partons inside colliding nucleons $N, N'$, and $l$ is lepton. The process $(1)$ describes the general process $N+N' \to l+l^{\dagger}+X$ in the case

$$

t_{\text{parton}}\gg t_{\text{int}}, \qquad (2)

$$

where $t_{\text{parton}}$ is a lifetime of the parton virtual state inside the nucleon, and $t_{\text{int}}$ is characteristic interaction time of the process $(1)$. The assumption $(2)$ is called impulse approximation and historically appeared in Drell and Yan papers on parton model.

In https://arxiv.org/pdf/1409.0051.pdf is written that on the modern QCD language $(2)$ is replaced by the (roughly speaking) factorization condition

$$Q^{2}\gg \Lambda_{\text{QCD}}^{2}, \qquad (3)$$

see Eq. (33), where $Q^{2}$ is the invariant pair of partons pair. The condition $(3)$ is simply interpreted as perturbativity condition in QCD. However, I don't clearly understand its relation to $(2)$, since the latter should obviously depend on the type of the interaction (as the interaction rate of the process, i.e. the right hand-side of $(2)$, is given by the transition probability per unit time).

Could anyone please explain why people use condition $(3)$ as the criterion of applicability of the Drell-Yan process for describing lepton pair production?