Spurious warning: too many AD fields

[RWilton]

When GATK tools HaplotypeCaller + GenotypeGVCFs report multiple alleles, one of which is a deletion, the AD field may contain an "extra" entry. This causes hap.py to emit a warning, "too many AD fields".

Some examples:

hap.py: WARNING  [W] too many AD fields at chr1:216200980 max_ad = 2 retrieved: 3
HC/GGVCF: chr1    216200981       .       C       *,A     122.04  .       AC=1,1;AF=0.500,0.500;AN=2;DP=11;ExcessHet=3.0103;FS=0.000;MLEAC=1,1;MLEAF=0.500,0.500;MQ=31.62;QD=20.34;SOR=3.912 GT:AD:DP:GQ:PL   1/2:0,3,3:6:99:381,132,152,132,0,156

hap.py: WARNING  [W] too many AD fields at chr9:111748759 max_ad = 2 retrieved: 3
HC/GGVCF: chr9    111748760       .       C       *,CCTT  158.01  .       AC=1,1;AF=0.500,0.500;AN=2;DP=6;ExcessHet=3.0103;FS=0.000;MLEAC=1,1;MLEAF=0.500,0.500;MQ=32.22;QD=26.33;SOR=2.303  GT:AD:DP:GQ:PL   1/2:0,2,4:6:51:224,168,162,58,0,51

hap.py: WARNING  [W] too many AD fields at chr5:116726417 max_ad = 2 retrieved: 3
HC/GGVCF: chr5    116726418       .       ATTATTTTATT     *,ATTATTTTATTTTATTTTATTTTATT    209.01  .       AC=1,1;AF=0.500,0.500;AN=2;DP=29;ExcessHet=3.0103;FS=0.000;MLEAC=1,1;MLEAF=0.500,0.500;MQ=38.62;QD=34.83;SOR=1.329  GT:AD:DP:GQ:PL  1/2:0,3,3:6:99:358,219,581,138,0,477

Over 10,000 instances of this warning may be generated using GRCh38 and a single 30x WGS sample.

I don't see anything in either the GATK docs or the VCF specification that explicitly describes the semantics of the AD field, but it seems to me that hap.py is emitting a warning when there should not be one.

Can you please help me sort this out?

Thanks!