Fix magnetization coupling

Model-specific changes:

The propagator MagnetosonicCurruntCoupling5D has been fixed.

In the old version, the Magnetization coupling term \nabla \times \iint f_h \mu \mathbf{b}_0 d v_\parallel d \mu \times \tilde{\mathbf{B}} was implemented as \sum_k^{N_p} \omega_k \mu_k \left\{ (\hat \nabla \times \hat{\mathbf{b}}^1_0) \times \hat{\mathbf{B}}^2\right\}(\eta_k).

It was replaced with \mathcal{T}_b^\top \mathbb{C}^\top \mathcal{P}_b^\top\sum_k^{N_p} \omega_k \mu_k \Lambda^0(\eta_k) .

\mathcal{T}_b is a time-dependent basis projection operator \tilde{\Pi} \left[ \hat{\mathbf{B}} \times \Lambda^\alpha \right] which depends on \tilde{\mathbf{B}}.

So the weights of the operator should be updated at every time steps.

*********** another approach to avoid projection into V⁰ space **************

• Projection operator \mathcal{P}_b is not used in here anymore, but instead accumulate \mathbf{b}_0 \cdot \Lambda² directly with the accumulation filling weight B^\mu_p = \omega_p * \mu_p * \frac{1}{\sqrt{g(\boldsymbol \eta_p)}} \hat{\mathbf{b}}¹_0(\boldsymbol \eta_p) \,.

So then now Magnetization coupling term in MagnetosonicCurrentCoupling5D step is \mathcal{T}_b^\top \mathbb{C}^\top \underbrace{\text{ACC}}_{\in \mathbb{R}^{N²}} .

This approach is also applied to another Magnetization coupling term in ShearAlfvénCurrentCoupling5D

• Codes are clean up based on the project documentation https://gitlab.mpcdf.mpg.de/struphy/struphy-projects/-/blob/main/running-projects/2022_DriftKineticCurrentCoupling.md?ref_type=heads

Documentation changes:

https://struphy.pages.mpcdf.de/struphy/sections/propagators.html#struphy.propagators.propagators_fields.MagnetosonicCurrentCoupling5D

has also been changed accordingly.