bug#13178: Query-replace fails soon after narrow-to-region

[Teika Kazura]

Docs of `use-empty-active-region' and `use-region-p' use the word "empty", but 
in fact the predicate is (> (region-end) (region-beginning)). It's confusing. 
(And also the Emacs Info, "Operating on the Region" section.)

For replace.el, the condition is correct (= what users expect.) Although I 
don't know other cases, maybe the explanation "empty" is too simplistic? (Sorry 
that I don't come up with a concise replacement description for that case.)

Thank you very much Emacs developers. And of course to Drew Adams for your kind 
reply and wide range, long time contributions.