Related paper from ghent and friends: https://arxiv.org/pdf/2404.14611

Could "relatively"(TM) easily implement a backend that supports abelian groups + fermionic grading (more precisely any symmetry with single fusion and symmetric braiding).

When the FusionTreeBackend is further along, we can compare the performance of FusionTreeBackend and AbelianBackend for some abelian symmetries and get a feeling if this abelian+fermionic backend what be relevant for performance.