Is it recommended to use voomWithQualityWeights() instead of conventional voom() when the data was already subjected to RUVg correction?

Applying RUV correction can sometimes clean up a dataset so well that quality weights are no longer required. In general though, the decision to batch correct and the decision to use quality weights are made separately. There is no general rule.

RUV is designed to correct for batch effects in the linear model. It is not designed to correct for differential variability associated with quality weights, so you should not necessarily assume that it avoids the need for quality weights.