parser/pageparser: Don't store the byte slices

On its own this change doesn't do any magic, but this is part of a bigger picture about making Hugo leaner in the
memory usage department.

parser/pageparser/pageparser_test.go

@@ -14,6 +14,7 @@
 package pageparser
 
 import (
+	"bytes"
 	"strings"
 	"testing"
 
@@ -88,3 +89,15 @@ func TestFormatFromFrontMatterType(t *testing.T) {
 		c.Assert(FormatFromFrontMatterType(test.typ), qt.Equals, test.expect)
 	}
 }
+
+func TestIsProbablyItemsSource(t *testing.T) {
+	c := qt.New(t)
+
+	input := ` {{< foo >}} `
+	items := collectStringMain(input)
+
+	c.Assert(IsProbablySourceOfItems([]byte(input), items), qt.IsTrue)
+	c.Assert(IsProbablySourceOfItems(bytes.Repeat([]byte(" "), len(input)), items), qt.IsFalse)
+	c.Assert(IsProbablySourceOfItems([]byte(`{{< foo >}}  `), items), qt.IsFalse)
+	c.Assert(IsProbablySourceOfItems([]byte(``), items), qt.IsFalse)
+}