Why don't # pages that are missing links + # pages that have links to target page = unique pages analyzed?

The seemingly missing pages aren't there because they didn't load. Typically these sites will simply be too slow to respond in the network timeout period set or the sites will simply no longer exist but still be in the index. The numbers (pages that are missing links and pages that have links to target page) are the total of pages actually loaded, rather than the number of urls returned by the SE.