Simplify the ED2 repository?

[mpaiao]

Context

I think the ED2 repository could use some trimming. It now has several unsupported features, and code that could go to a separate repository. I think this would make the code lighter and easier to maintain. I was the one who added most of the stuff I suggest removing, but I know a few people may be using some of these, so I would like to hear from them.

Proposed changes

I think the following should be removed completely:

1. RAPP. This was a pre-processor to convert NCEP reanalyses to HDF5 that I wrote back in the last glacial maximum. I don't think anyone is using it or supporting it. I have R-based scripts that convert reanalyses to the ED2 format in this repository, which are more up-to-date.
2. BRAMS and Ramspost. I don't think anyone is using ED2-BRAMS, so I think we could make a final tag with ED2BRAMS, and delete both BRAMS and Ramspost (BRAMS post-processor). Some of the code may still be useful if we plug ED2 into ESMs, but we can borrow the code from the last version. The coupler should likely be hosted by the ESMs in any case, not in ED2.
3. R-utils. These are a collection of functions that I wrote over the years, and many are not ED2-related. This could be migrated to a separate GitHub, along with additional R scripts (see below).
4. ED/Template. These are some utility scripts for launching sets of ED2 runs. These are rather specific for the Harvard Cluster (and some clusters at LNCC), and likewise could go to a separate repository.
5. ED/src/preproc. This is not something I added. My impression is that no one is using it, and these are superseded by the scripts available in this repository.
6. ED/test_cases. This is not something I added. I think these may have been part of an early version of the test suite, but since then they have been superseded by the continuous integration (which could use a couple more tests for the tropics and edge cases like hyperarid deserts).

In addition, I think EDTS should be slimmed down to be just the files needed for the continuous integration. I love the test suite idea, but most of the code is in Matlab and no one has been really using it for years. Some elements of the test suite could be eventually re-integrated as part of the continuous integration using some free software and a much simpler structure.