Documentation regarding the loops over groups in the data structure.

[semi-h]

@JamieJQuinn pointed out in #67 that its not clear for the user how to specify loops over groups.

I think we need a section in https://xcompact3d.github.io/x3d2/internals/ describing how we loop over groups in the domain both in CUDA and OpenMP backends. This is closely related to #68 and the subsequent changes that'll come in CUDA backend regarding reorderings as well.