RGBA8888_LE written in wrong byte order to framebuffer

While investigating BIG-endian issue in MIPS QEMU environment, a typo in a byte order in referenceBlit() and few other places (in tst_qsaferenderer.cpp) was found out. As runtime was tested against wrong byte order the convertAndBlit() function in qsaferenderer.cpp had wrong byte order in RGBA8888_LE case. The byte order happened to be BGRA instead of ABGR as intended, i.e. the alpha was in wrong place.