Fix a performance bug where using -w could result in very bad performance.

The specific issue is that -w causes the regex to be wrapped in Unicode
word boundaries. Regrettably, Unicode word boundaries are the one thing
our regex engine can't handle well in the presence of non-ASCII text. We
work around its slowness by stripping word boundaries in some
circumstances, and using the resulting expression as a way to produce match
candidates that are then verified by the full original regex.

This doesn't fix all cases, but it should fix all cases where -w is used.

grep/src/lib.rs

@@ -19,6 +19,7 @@ pub use search::{Grep, GrepBuilder, Iter, Match};
 mod literals;
 mod nonl;
 mod search;
+mod word_boundary;
 
 /// Result is a convenient type alias that fixes the type of the error to
 /// the `Error` type defined in this crate.