Read length distribution depends on sample quality, putative contaminants (see e.g. https://community.nanoporetech.com/contaminants), and for the amplicon product of course on the PCR specificity. While an agarose gel is definitely a good way to check the result of the PCR it lacks in sensitivity to detect bands with a low abundance (e.g. in comparison to electropherograms of a Bioanalyzer or Fragment Analyzer). These are usually the main reasons of “abnormal” read lengths distributions.
Most of the samples have read lengths cut off at the assembled size, which is a good indication that the full-length amplicon was assembled (and they also have coverage plots with an even read depth/coverage)