I guess this is not only specific to DESeq2 but also to other methods like edgeR, etc.

I do not have any data for this or any code. It is more of a general question.

Assume you have the following setup: - 5 Control samples - 5 Treated samples

Now, also assume that we have a rather "low effect" from the treatment on the expression of genes. Would you be surprised if you find more sign. DE expressed genes if you take only 4 treatment samples compared taking all of them (5 treatment samples). Or different: Is it a common effect to get less sign. DE genes if you have a higher sample size? Again, especially if you do not have strong effects from the treatment.

You would generally expect more rejections of non null genes with higher sample size. However, its easy to imagine situations where removing a particular sample or set of samples (eg failed samples) where you recover more signal without them.