Lots of improvements. Most notably, removal of memory maps for searching.

Memory maps appear to degrade quite a bit in the presence of multithreading.

Also, switch to lock free data structures for synchronization. Give each
worker an input and output buffer which require no synchronization.

src/printer.rs

@@ -28,6 +28,10 @@ impl<W: io::Write> Printer<W> {
         wln!(&mut self.wtr, "{}", path.as_ref().display());
     }
 
+    pub fn path_count<P: AsRef<Path>>(&mut self, path: P, count: u64) {
+        wln!(&mut self.wtr, "{}:{}", path.as_ref().display(), count);
+    }
+
     pub fn count(&mut self, count: u64) {
         wln!(&mut self.wtr, "{}", count);
     }