Improve oshash code (#1829)

Reduce allocations. Don't create intermediary arrays which we then
consume right after. Manually fuse the arrays and decode straight into
the sum instead.

Furthermore, don't invoke a Reader, but carve out the locations via a
loop, directly.

These two changes taken together speeds up oshash computations by a
factor of 10 according to the benchmark tests. The main reason for
this change is a much lowered memory allocation rate which in turn
improves GC pressure.

While here, add a benchmark for oshash computations and use it for
testing the performance.

pkg/utils/oshash_internal_test.go

@@ -1,6 +1,7 @@
 package utils
 
 import (
+	"math/rand"
 	"testing"
 )
 
@@ -44,3 +45,31 @@ func TestOshashCollisions(t *testing.T) {
 		t.Errorf("TestOshashCollisions: oshash(n, k, ... %v) =! oshash(n, k, ... %v)", buf1, buf2)
 	}
 }
+
+func BenchmarkOsHash(b *testing.B) {
+	src := rand.NewSource(9999)
+	r := rand.New(src)
+
+	size := int64(1234567890)
+
+	head := make([]byte, 1024*64)
+	_, err := r.Read(head)
+	if err != nil {
+		b.Errorf("unable to generate head array: %v", err)
+	}
+
+	tail := make([]byte, 1024*64)
+	_, err = r.Read(tail)
+	if err != nil {
+		b.Errorf("unable to generate tail array: %v", err)
+	}
+
+	b.ResetTimer()
+
+	for n := 0; n < b.N; n++ {
+		_, err := oshash(size, head, tail)
+		if err != nil {
+			b.Errorf("unexpected error: %v", err)
+		}
+	}
+}