On the rate of in-reactor UO2 densification