CC->CC transition for general models fails when n_mesh_p is not the same for each kernel

[levisc8]

It looks like for at least 1 model in Padrino, evaluation of the expressions produces an output of the wrong length, which breaks .fun_to_iter_mat. This is because each domain (n_seedling = 137, n_ramet = 500) aren't properly crossed. This is not an issue when n_seedling = n_ramet.

Need to implement crossed versions of these domains for models where there are multiple continuous states, and then work out how to replace these internally, as the user-facing API shouldn't change.