That's mean it as an ASV table, not collapsed to species level.
Here you can read how to add taxonomy to the biom file or to collapse feature table.
Now I undesrtood - you mean that each ID should be unique, not the counts of one ID in feature table among your samples. It is right. But still, you are working with ASV table now, and they are not species. If you collapse your table to species level, each species should be unique in your new collapsed table.
Rarefaction is for subsampling, not for taxonomy assignment.