You're right that the main reason for choosing a subsample for some biomarkers in some DHS surveys is to keep the sample to a manageable size. Biomarker tests can be time consuming and expensive, and they have the potential to distract attention from the interviews. Since the subsamples are always based on a random selection of households, the biomarker results are based on a random sample of households and individuals, even though the ultimate sample size will be reduced.