max_mhd_level=0 default silently breaks pure-MHD runs

[nicolasaunai]

Why

max_mhd_level defaults to 0 (check_max_mhd_level, pyphare/pharein/simulation.py:658-666), which makes sense for the Hybrid default case (model_options="HybridModel", no MHD levels at all). But nothing stops you from doing Simulation(model_options=["MHDModel"], ...) and leaving max_mhd_level untouched, and in that case things don't just behave oddly, they break silently.

In src/simulator/simulator.hpp:358:

multiphysInteg_->registerModel(0, maxMHDLevel_ - 1, mhdModel_);

With maxMHDLevel_ = 0 this is registerModel(0, -1, mhdModel_), an empty range, so no level gets the MHD model registered. The Hybrid-model registration branch right above it only runs when model_options includes "HybridModel", which it doesn't here. End result: a model_options=["MHDModel"] run with the default max_mhd_level ends up with literally no physics model attached to any level in the hierarchy.

What

Either:

• error out at Simulation construction time if model_options=["MHDModel"] and max_mhd_level is left at 0, or
• default max_mhd_level to max_nbr_levels when model_options is MHD-only, since that's what "every level is MHD" (the existing doc note for a pure MHD run) actually requires.

Ideas on how (optional)

In check_max_mhd_level (pyphare/pharein/simulation.py), make the default depend on model_options instead of being a flat 0:

def check_max_mhd_level(**kwargs):
    default = kwargs["max_nbr_levels"] if kwargs["model_options"] == ["MHDModel"] else 0
    max_mhd_level = kwargs.get("max_mhd_level", default)
    ...

(need to check ordering vs. check_model_options in checker, since model_options is currently validated after max_mhd_level)

Alternatives considered

No response

Additional context

Found while documenting max_mhd_level in the Simulation docstring.

[nicolasaunai]

Also, the name itself is misleading regardless of the default. In simulator.hpp:358, registerModel(0, maxMHDLevel_ - 1, mhdModel_) registers levels 0 through maxMHDLevel_ - 1 as MHD, so max_mhd_level is really a count of how many levels (starting from 0) are MHD, not the index of a level. max_mhd_level=2 means "2 levels are MHD", not "level 2 is MHD".

That's inconsistent with max_nbr_levels, which is also a count and is named accordingly. max_mhd_level should probably be renamed to something like max_nbr_mhd_levels (or nbr_mhd_levels) to match.