The only item that came up that I'm unable to address myself is discussion comparing a NOT NULL column constraint to an equivalent check constraint. I've left things documented as semantically equivalent. A future patch can clear those things up. At this point I'm considering this patch content complete.
I figure to make any last tweaks against this version 7, combine the two patches into one and submit v8 as ready to commit should a reviewer agree.
For now I've left v6 alone and added a diff to cover these last changes.
The markup surrounding the examples is correct now and I decided \N was the most useful representation of NULL given that the query data is single digit numbers. I really hate the non-readability of t/f output for booleans so I manually cleaned those up.