Several updates to Variational models to run itpa test cases

Core changes:

in BasisProjectionOperator : Allow to pass the Boundary projector and Polar extractor for the space of P by hand, to handle cases where one might not want to apply to boundary operator there.

in MassMatrixDiagonalPreconditioner take the temporaries in the space of the matrix and not the WeightedMassOperator as this might be the polar space.

in L2Projector add a special case for Polar vector (as one need to use vector._tp_data instead of directly vector._data)

same in the evaluation methods for Fields in the Derham class

add a parallel method to PolarDerhamSpace as this might be needed when building block operators, and remove direction argument from update_ghost_regions as this is not recognize by psydac. Also add an exchange_assembly_data method, required by psydac when using BlockLinearOperators.

Model-specific changes:

Removed VariationalBurgers model as well as VariationalVelocityAdvection propagator as they were not used, optimized and since the propagator was not part of the VariationalMHD model I judged useless to keep it.

Update variational_utilities to take into account possible polar splines and boundaries.

Update all the Variational models and propagator to take into account possible polar splines and boundaries.

Update the s0 method on MHDequilibrium to not rely on p0.

Add the p0 option to the AdhocTorus equilibrium, this will add a constant to the pressure to avoid pressure dropping to 0 at the boundary, causing problem with the entropy going to minus infinity.

Add an optional parameter to TorusModesSin with pfuns = sin, allowing to have other modes then 1 for the prefactor.

Change the value of the form used in the approximation of the advection operator (now use an implicit approximation for more stability).

Documentation changes:

Update the docs with this new "implicit" transport operator.