Aha.. You have a mixed bag of charts- not all new ones from the CHS? and including Polyconic charts. It's got to be hard to mesh different projections ....even for OCPN.
As I comprehend it, Polyconics were an especial way to make a harbour appear on the chart more naturally from the wheelhouse versus a mercator, so that landforms can look like they do in reality, versus Mercators that distort appearance more and more in higher latitudes . Lots of old surveys in North America used it, on small
areas. But you can look it up yourself and see what conclusion the internet
Meanwhile, an easy fix might be to investigate and create "Chart Groups" from the OpenCPN manual and group the Polyconics...
The problem of laying a sphere onto a flat surface has entertained generations of chart and map makers and still does. When these charts were surveyed these charts were made to print and to be used singly, on a table with a pencil. I think "quilting" (while not unkown) was not a requirement and still is not.
ADDed: just surmise, but if you have a chart covering a harbour, you want the whole chart to be accurate, including the edges where the approaches to the harbour may be. Perhaps a polyconic satisfies this, where a mercator does not.