When I suggested this connection I did consider Bandiagara but left it out on a"judgement call". Main reasons against - It isn't a trek with a standard route (or routes) or final "walking" objective (which could be in the form of to "reach A" or "circle B").
Els mentions that it isn't a trek you do for the "walking aspect" but then I suppose neither are The Routes of S de C! I wouldn't argue against its inclusion given its popularity (whatever "it" is)
The SW coastal path actually seems to hit more of the right buttons. It IS a walk with a specific "name" and a defined start/finish. A potential problem is that many countries are building up a portfolio of "long distance paths" - in Europe there are even transboundary paths with "E" numbers crisscrossing the continent
http://en.wikipedia.org/wiki/European_long-distance_paths and in France of course the Grandes Randonees! This site has a worldwide list
http://en.wikipedia.org/wiki/List_of_long-distance_footpaths So where do we stop- Just the "E routes" are likely to go through vast numbers of WHS. I feel that the path needs to be of a size which is likely to be done "in full" by people (OK there is probably someone who at this very instant is nearing a completion of all the E routes -there are always such exceptions (I understand that there are even people who collect WHS sites!) but that would not be the norm unlike eg The Inca Trail.
So, among the selection criteria I would add the following for consideration.
a. A specific name
b. Defined route or routes
c. Often/normally done in total