Add option for deep atmosphere

[Sbozzolo]

This is a rebased and cleaned (from a git perspective) version of #1591 (authorship is preserved).

This PR adds supports for deep atmospheres and changes a little bit the hypsography interface.

I am not particularly satisfied with the current hypsography interface, in particular in relation with the interpolation methods.

Among the problems are:

• Converstions between scalars and ZPoints. The interpolation interface works with scalars, the hypsography with ZPoints, so there are conversions that need to be done between the two
• Restarting from HDF5 currently goes through branches that are hardcoded for the hypsographies we support, but there is no reason why this should be the case.
• At the moment, both the "reference to physical z" and "physical to reference z" transformations are required, but some coordiante transformations do not have closed forms for both (e.g., our SLEVE), so that some features in the interpolation interface cannot be easily implemented.

In the interest of merging support for deep atmosphere, we can go ahead with this PR and tackle the issues with hypsography/interpolation in the future.

Closes #1524

[akshaysridhar]

Thanks @Sbozzolo. This is ready to rebase and merge with the included changes against commit c81d857.

[charleskawczynski]

I left a few more comments, but overall looks good.

[akshaysridhar]

I left a few more comments, but overall looks good.

Thanks @charleskawczynski ! My opinion is that the questions about Point support be addressed in future PRs related to diagnostics / interpolations (type conversion support etc.) . I have merged your suggestions w.r.t dss bufers and alloc reduction in the smoothing function.

[Sbozzolo]

I rebased, squashed, and addressed the comment on map. I don't know the answer for ZPoint. Good to go on my end.