Update and add more examples and documentation

[bgroenks96]

Following #68, we will need to do a major overhaul of the documentation, which is anyway now several months out-of-date.

This should be a top priority since it's a major issue blocking a v0.1 release.

I think this can be broken down into a few subtasks:

• Update documentation on AbstractModel and AbstractProcess interface
• Update existing documentation on physics to include direct references to the relevant AbstractProcess types and coupling interfaces.
• Add documentation on closure! and invclosure! interface as well as closure relations generally.
• Add documentation on process coupling (probably in a new developer guide section)
• Add example for soil hydrology
• Improve existing example for coupled vegetation-soil model

[maximilian-gelbrecht]

I'll make a first draft of an updated example_model_notebook.jl today.

But you are absolutely right in saying that this is a top priority. Especially if we want to submit to JOSS very soon, the documentation is what reviewers and editors look at the most.

[maximilian-gelbrecht]

But before we document everything...

Are we happy with the structure now or do we still consider moving the kernel launching? That's something we talked about. I think it could be okay now, but maybe we should have a quick think and benchmark to be certain

[bgroenks96]

What do you mean by moving the kernel launching? Do you mean adding more kernel fusion? Because yes, we should do that, but it's not a fundamental change to the current structure, it would just be adding new kernels to the coupled process types.

[maximilian-gelbrecht]

We already talked about it, it was rather about kernel fusion and where to do it or not, but I see that it wouldn't change the overall structure.

[bgroenks96]

The only part about the structure/API that I am unsure about is the decision in the last refactoring to make the processes all have varying numbers of arguments based on their dependencies. Of course I still support the reasoning behind this that it makes relationships between processes more clear, but I am afraid that it may come back to haunt us as more processes get added.