It's hard to tell what the 12Mb of sequence they are calling "other" is. It could be bacterial, or a decoy contig that is used to capture low-quality reads. Sometimes you'll see reference sequences also include things like EBV. Or it could be the ALT assemblies that represent different ways of putting together the reference (like HLA differences).
The fact that you got 39X raw reads is pretty good.
Also, it is usually called hg38, not hs38... I'm not sure where that came from. If you want to be super accurate, you could also use GRCh38, but there are sometimes subtle differences between the two.
https://gatk.broadinstitute.org/hc/en-us/articles/3600358909...
Eh, your results look perfectly fine, especially for a budget sequencing run. Keep in mind there are a lot of repetitive DNA sequences that don't map well (search "mappability track" for more info). Depending on the reference they used, these may even be masked out and thus no alignment can even be attempted.
My current best guess is that more data than usual were of bacterial origin, but I'll have to learn a bit more bioinformatics to test that.