Simple LES and RANS models added with `extend_simul_class`

We want to be able to do something like:

from fluidsim.solvers.ns3d.solver import Simul
from fluidsim.base.turb_model import SmagorinskyModel, extend_simul_class

Simul = extend_simul_class(Simul, SmagorinskyModel)
params = Simul.create_default_params()
params.turb_model.enable = True
params.turb_model.type = "smagorinsky"
params.turb_model.smagorinsky.C = 0.18

...

sim = Simul(params)
sim.turb_model

Here, we want to limit ourselves to simple models which only change the non linear terms and do not need other solved variables.

I guess we need to modify a bit fluidsim/base/solvers/base.py and the solvers (for example fluidsim/solvers/ns3d/solver.py) (see !308 (merged)).

I first wrote:

        if self.is_turb_model_enabled:
            tendencies_fft += self.turb_model.get_model()

but we should define get_model as

    def get_model(self, **kwargs):
        ...

and give some variables already computed/available in tendencies_nonlin.

There should be a class for the sim.turb_model object (very similar to the equivalent class for forcing) and classes for specific turbulence models (very similar to SpecificForcing).

I don't know if we need to compute the turbulent model for each call of tendencies_nonlin or just once per time step (as for the forcing).