Context Navigation

Changes between Initial Version and Version 1 of Ticket #14654, comment 26

initial	v1
1	1	Hmm. I don't like the idea that the sources and the sinks are inferred. I'd prefer to have the user provide them (and letting the code infer them only if so required), as I can see myself applying such a method for various sink-source configurations on a single graph, and it looks strange to build a new graph every time.
	2
	3	(I was planning to make these changes myself soon enough, once I'm done with Gelfand-Tsetlin and Berenstein-Zelevinsky polytopes, but I caught something last week and now all I'm swamped with work.)