Use into_iter instead of iter.

It's slightly faster. On a huge file, old:
```
Benchmark 1: target/release/counts big2 > /dev/null
  Time (mean ± σ):     942.4 ms ±  21.1 ms    [User: 884.6 ms, System: 57.7 ms]
  Range (min … max):   917.7 ms … 971.3 ms    10 runs
```
and new:
```
Benchmark 1: target/release/counts big2 > /dev/null
  Time (mean ± σ):     932.6 ms ±  14.2 ms    [User: 874.2 ms, System: 58.4 ms]
  Range (min … max):   902.8 ms … 952.0 ms    10 runs
```

Author: nnethercote

src/main.rs

@@ -158,15 +158,15 @@ where
 
     // Sort from highest count to lowest count. For lines with the same count,
     // sort them in alphabetical order.
-    let mut counts: Vec<_> = counts.iter().collect();
+    let mut counts: Vec<_> = counts.into_iter().collect();
     counts.sort_unstable_by(|(line1, n1), (line2, n2)| {
         (n2.abs(), line1).partial_cmp(&(n1.abs(), line2)).unwrap()
     });
 
     writeln!(io::stdout(), "{:.1} counts{}", total, label)?;
     let mut cum_perc: f64 = 0f64;
     let total_f64 = total.into_f64();
-    for (i, (line, &weight)) in counts.iter().enumerate() {
+    for (i, (line, weight)) in counts.iter().enumerate() {
         let perc: f64 = weight.into_f64() * 100f64 / total_f64;
         cum_perc += perc;
         writeln!(