Radiative module bottleneck
In problems with full radiative energy treatment the code spend about 80% of the time in the SOR solver.
the SOR is called at every dynamical timestep, there is no control on the timestep from the energy equation.
The radiative diffusion timescale however can be very different from the dynamical timescale. So one could have a dynamical (dt) and an "energy" timestep (with dtEne >dt) and call the radiative routine only every dtEne
@astr-mle :this would be particularly interesting when using the non-uniform grid since in this case we can't use the FARGO algorithm and the dynamical dt may be very small (~0.001)
This issue and related branch are devoted to the introduction and test of dtEne