I have found a dataset where the 16rRNA amplicon size is 292 bp while the sequencing reads are 300 bp. Is it possible or is it a mistake by the people who uploaded the data in NCBI?
You are talking about Sanger sequencing or output of any NGS platform? You can check the quality of those reads. If the quality score dropped abruptly at the end of amplicons length (292 bp in your case), then there is chances that erroneous bases have been incorporated. In this case, you can trim those bases and go ahead with high-quality reads.