Revisiting/improving the "ecritphy" parameter to specify output rate in physics

[emillour]

In the current setup (for Mars/Generic/Titan?) PCMs, the output rate for "diagfi.nc" files is specified via the "ecritphy" parameter (in run.def)
But ecritphy has to be given in dynamical time steps, which can be a bit confusing because then it has to be a multiple of "iphysiq" (and also this is for LMDZ.COMMON and not very convenient when switching to DYNAMICO...)
It would probably be better to be able to specify physics output rate in terms of physic time steps, independently from the dynamical time steps and rate of call to the physics.

For the Mars PCM this was initiated with r3369, with the choice of introducing an "outputs_per_sol" flag (to be used instead of ecritphy).