GATK 3.8: genotype called as HOM_VAR while there is a large number of REF and only a few ALT

Hi the GATK team,
there is a mutation in my (haloplex) bam that was called (REF=C) as HOM_VAR in my bam while there are only a few reads carrying a ALT=A

$ samtools mpileup -r "15:42171531-42171531" ~/jeter.bam 
[mpileup] 1 samples in 1 input files
15  42171531    N   735 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc$cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccaacccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccacccccccccccccccccccccccccccccccccccccccccccccccccccCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC    

I put a minimal example in https://nextcloud-bird.univ-nantes.fr/index.php/s/B4WQSw3oLwsWwYS

HaplotypeCaller (3.8-0-ge9d806836) in vcf or g.vcf mode ( I'd like to stick to 3.8 for now )

my command line:

java -Djava.io.tmpdir=. -jar ${GATK_JAR} -T HaplotypeCaller -R "/path/to/b37/human_g1k_v37.fasta" -I ~/jeter.bam  -L "15:42171520-42171550"
(...)
15  42171531    .   C   A   2080.77 .   AC=2;AF=1.00;AN=2;BaseQRankSum=-3.799;ClippingRankSum=0.000;DP=90;ExcessHet=3.0103;FS=12.355;MLEAC=2;MLEAF=1.00;MQ=60.00;MQRankSum=0.000;QD=24.19;ReadPosRankSum=1.079;SOR=6.982    GT:AD:DP:GQ:PL  1/1:5,81:86:99:2109,133,0

here the genotype is 1/1 AD=5/81 DP=86

Furthermore, the DEPTH looks much higher with samtools:

$ samtools depth -r "15:42171531-42171531" ~/jeter.bam      
15  42171531    833

what could be the cause of this ?

thank you in advance for your help,

Pierre