Neuron to Graph: Interpreting Language Model Neurons at Scale