In GenomicsDBImport and GenotypeGVCFs, we use single-sample GVCF files as input and there are annotation fields for single sample in each GVCF like BaseQRankSum,ClippingRankSum and ReadPosRankSum. How does GenotypeGVCFs combine annotation values (e.g., BaseQRankSum) from individual GVCFs to generate population-level metrics when performing joint genotyping? Is it like  what we do in a meta-analysis?