CLIPTER: Looking at the Bigger Picture in Scene Text Recognition