Address memory leak in SkScaledImageCache

This is intended to address the valgrind complaint:

1,864 (128 direct, 1,736 indirect) bytes in 1 blocks are definitely lost in loss record 823 of 916
==4572==    at 0x4C2B1C7: operator new(unsigned long) (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==4572==    by 0x55072A: SkScaledImageCache::addAndLock(SkBitmap const&, float, float, SkBitmap const&) (SkScaledImageCache.cpp:468)
==4572==    by 0x440791: skiatest::ImageCache_doubleAddClass::onRun(skiatest::Reporter*) (ImageCacheTest.cpp:121)
==4572==    by 0x48C07A: skiatest::Test::run() (Test.cpp:107)
==4572==    by 0x4863D0: SkTestRunnable::run() (skia_test.cpp:155)
==4572==    by 0x629D95: SkThreadPool::Loop(void*) (SkThreadPool.cpp:97)
==4572==    by 0x634462: thread_start(void*) (SkThreadUtils_pthread.cpp:66)
==4572==    by 0x4E39E99: start_thread (pthread_create.c:308)
==4572==    by 0x6A02CBC: clone (clone.S:112)

[reed1, 2013-12-18 18:43:27]

lgtm

[robertphillips, 2013-12-18 18:48:30]

committed as r12749