Input for dynamical interactions

[andrewkhardy]

I am sure this on the TRIQS developer's radar but a few points I think that are worth checking for v4.0.

1. The input for the dynamical interactions for CTINT appear to be in frequency (there is a Solver.Jperp_iw object but no Solver.Jperp_tau.) I think to bring this into alignment with CTSEG and (soon) CTHYB, the inputs should be in $\tau$, particularly since these inputs will likely be $\tau$ grid measurements.
2. The other two solvers infer use_Jperp and use_D from the inputs. This seems like a worthwhile improvement to align, of if there is a good reason to have a separate flag, the other two solvers should as well.
3. Small aside, if the solvers are to have DLR grids as the default, I also think the default for fit_gf_dlr should be have the same bool for symmetrize as the solver, likely symmetrize=True.