Automating and Simplifying MLPerf Inference Benchmark Submissions with TVM and CK