move mutex guard from inside a loop (#29)

# What ❔

This PR is implementing a small performance improvement in the device
allocation code by moving a mutex guard form inside a loop to the
outside.

## Why ❔

Acquiring mutex in a loop is suboptimal.

## Checklist

- [x] PR title corresponds to the body of PR (we generate changelog
entries from PRs).
- [x] Code has been formatted via `cargo fmt` and linted with `cargo
check`.

src/static_allocator/device.rs

@@ -251,8 +251,12 @@ impl StaticDeviceAllocator {
         None
     }
 
-    fn free_block(&self, index: usize) {
-        self.bitmap.lock().unwrap()[index] = false;
+    fn free_blocks(&self, index: usize, num_blocks: usize) {
+        assert!(num_blocks > 0);
+        let mut guard = self.bitmap.lock().unwrap();
+        for i in index..index + num_blocks {
+            guard[i] = false;
+        }
     }
 
     pub fn free(self) -> CudaResult<()> {
@@ -351,10 +355,7 @@ unsafe impl Allocator for StaticDeviceAllocator {
         assert_eq!(offset % self.block_size_in_bytes, 0);
         let index = offset / self.block_size_in_bytes;
         let num_blocks = size / self.block_size_in_bytes;
-        assert!(num_blocks > 0);
-        for actual_idx in index..index + num_blocks {
-            self.free_block(actual_idx);
-        }
+        self.free_blocks(index, num_blocks);
         #[cfg(feature = "allocator_stats")]
         self.stats.lock().unwrap().free(index);
     }