The method HyPerConn::convertPreSynapticWeights produces incorrect results near borders and fails to run if pre and post layers are of the same size. Also won't run if pre-synaptic layer is denser (more neurons) than post.
