PCR of repeats is used to measure their length, which is highly polymorphic. But since Taq ads additional A at the end in presumably (from the paper) unpredictible manner, and together with the fact that polymerase may slip on the repeats, it then creates multiple products from one reaction, differing around that one nucleotide.
Products are marked by fluorescent fwd primer and analysed in capillary electrophoresis (similar as sequencing). But the you got multiple signals for one length and if you are looking into dinucleotide repeats for example, it's really difficult to measure the right length then.
This is for the explanation why this is needed.
This seems to be the original paper
about the added A.
Maybe more will be found in key papers about STR polymorfisms, like this one maybe, but I don't have the access. Validation of short tandem repeats (STRs) for forensic usage: performance testing of fluorescent multiplex STR systems and analysis of authentic and simulated forensic samples.
But from search seems, that adding tails (not just this particular sequence) is a common thing in STR amplification.
Keep me posted if you find out