cluster density
 

Is it possible that a high cluster density >1300K/mm^2 that may result in difficulties for the MiSeq software to distinguish/misidentify neighboring clusters to intorduce errors i.e. the sequence diverge (maybe due to SNV or indel) and the software cannot identify the correct sequence and this results in a frame shift call?

If 20 reads, from 20 different clusters, going in both directions and all starting from different positions in the genome, and all say there's a frameshift, you have a frameshift. It would be a mistake to conclude anything from a single read, if that's what you are asking.