Currently TensorCast (or TransmuteDims in particular) requires Strided v < 2. This makes the DMRG incompatible with the current package (TBT as well). As those packages seem less well maintained to begin with, we should just switch everything over to ITensors.jl
Since we're hoping to build this large-N solver on top of MFT, we should take the time to get proper Matsubara integrators and everything. This means adopting MatsubaraFunctions Even in real frequency, it would be nice to have better adaptative frequency integrators.