All the numbers are there in the popup, there is nothing hidden. 3,949 reads have "C" in the read sequence at that location, 1 has "A", 41 have "N", and 938 have a gap (deletion). So there are 3,991 + 938 = 4,929 reads spanning that locations. Insertions are special because they are really between locations, or to one side so-to-speak, by definition they do not have any genomic location. So it would be incorrect to add the insertions to the total pileup. Put another way, imagine a single read with an "A" at a location, and an insertion adjacent. You would not have 2 reads spanning the location, only 1.