We should simplify the MPOs for the hybrid real/momentum space DMRG benchmarks (https://github.com/ITensor/ITensors.jl/blob/master/examples/src/hubbard.jl#L71-L79 and https://github.com/ITensor/ITensor/blob/v3/sample/src/hubbard.h#L45-L54) and rerun the benchmarks with the simpler MPOs, now that https://github.com/ITensor/ITensors.jl/pull/519 is fixed. Note that we also need to fix #1 before rerunning.