You may actually be fine if you actually write a program to inject code rather the redistributing DF with your changes in it, but it still doesn't hurt to read it at least. Granted, this sort of worry is still a long way off. I was just curious myself he'd actually put in his copyright so I went and looked it up.
I never intended to redistribute DF - not sure where you got that idea from. I will write the thing and release it as dfhack/foreman et. al. are released. Code injection will happen at the runtime - it's possible to do it with a running process.
You may want to look at this thread, specifically this simulator someone already wrote. When we say that pathfinding has been discussed extensively, we're really not kidding.
This link is interesting and I bookmarked it for a reference for the future when I do the final optimisation of the pathfinder. Cursory glance at it does indicate that the guy doesn't have a right heuristic, for the DF. He should add a diagonal distance (one outlined @ http://theory.stanford.edu/~amitp/GameProgramming/Heuristics.html
will work for 2d map). I derived a formula suitable for df from this - it's important to have H and G in balance, otherwise you don't get the shortest path or the pathfinding will be slow.
Anyway I have a workable pathfinding algorithm implemented - caching the paths and proper invalidation of the cache will be the hardest part - after all, cache invalidation and naming things are two most difficult problems in programming;)