bu in Path.Relativize

[Alan Bateman]

Rémi Forax wrote:
> :
> Thank you, Alan,
> a workaround is to used realPath() on both paths before calling 
> relativize.
I've created 6925169 for this and I have an initial fix that I'll send a 
webrev soon. While looking at this I see that  we've got a subtle spec 
bug in p.relativize(p.resolve(q)).equals(q) can only be guaranteed to be 
true if p and q are normalized paths. I'll fix that too.

-Alan.