I mistakenly thought that the other GPS I have connected had the same delays, but I was looking at the same graph by mistake! I see now that GPA.DELAY has the expected 200 mSec +/- 20 mSec very perfectly. Only the M8P (above on GPA2.DELAY) had the wild swings. Attached is the well-behaving GPS graph, which is a Ublox M8N - NO problems with it
.